Scribe v3
Scribe v3 and Scribe v3 Live, the most accurate ElevenLabs speech recognition models, are now available in the Speech to Text API. Scribe v3 (scribe_v3) transcribes recorded audio, and Scribe v3 Live (scribe_v3_live) transcribes in real time. They succeed Scribe v2 and Scribe v2 Realtime, which remain available.
Scribe v3 makes 27% fewer word errors than Scribe v2 on the FLEURS benchmark across 102 languages and labels speakers more accurately. It adds numbers_format to write numbers as digits or words, and secondary_language_codes for audio that mixes languages. Scribe v3 Live has a 44% lower semantic word error rate than Scribe v2 Realtime on FLEURS across 73 languages. It adds turn prediction, which commits a segment when the speaker has finished their turn instead of at the first pause, a multilingual mode that writes each language in its own script, and a keyterm library of up to 10,000 terms.
Unity SDK
The ElevenAgents Unity SDK is now available as the io.elevenlabs.agents package, starting with v0.1.0. It supports the Unity Editor, desktop standalone players on macOS, Windows and Linux, and WebGL builds, with iOS, Android and XR in beta. It requires Unity 6.3 LTS or later.
The SDK connects a Unity scene to an agent through a single Conversation API. It captures the player’s microphone, plays the agent’s voice through an AudioSource you can position in the scene for spatial audio, and raises transcript, response, mode and tool call events on Unity’s main thread. Desktop and mobile builds connect over WebSocket, and WebGL builds can also use WebRTC.
Workflows as code
ElevenAgents workflows can now be built in code with the Workflow builder in the Python and JavaScript SDKs, from v2.77.0. Chain converse, tool, transfer and end steps with when, on_success and go_to routing, then pass the result as workflow to agents.create() or agents.update(). The builder runs entirely in the SDK, so you can load an agent’s workflow, change one node, keep workflows as JSON in version control and still edit them in the visual editor. See Workflows as code.
More agent tools in the hosted MCP server
Hosted MCP adds 24 tools for inspecting agents, covering analytics, configuration summaries, insights, alerts, test summaries, branch conflicts and conversation counts. They use the existing convai_read and convai_write permissions, so connected MCP clients can review an agent the way Architect does. MCP clients can also find and replace or append text in an agent’s system prompt or first message, staged in a branch draft until it is published. See Hosted MCP.
All changes
ElevenAgents
- Workflows as code: The Python and JavaScript SDKs add a
Workflowbuilder for agent workflows. See Workflows as code. The workflow editor adds a minimap, zooms out further and has a snap-to-grid toggle, and exposes the subagentturn.soft_timeout_configoverride so dashboard saves no longer remove it. - Clone strength and reasoning effort overrides: Agent
tts.clone_strength,clone_strengthon supported voices and aclone_strengthconversation override are now in the public API. Clients can setconversation_config_override.agent.prompt.reasoning_effortonce the agent owner enables the Reasoning effort override; a level the agent’s LLM does not support stops the conversation from starting. Realtime agents show the stability, similarity and clone strength sliders. See Overrides. - Conversation data: List conversations returns
cost(credits) andcost_fiat(USD), and the conversations table has an optional Cost column. Transcript rows on asynchronous channel conversations addcreated_at_unix_ms, and asynchronous conversation metadata addsintegration_trigger_idandplatform_metadata, such as the Zendeskticket_id. Post-callaudioand unredacted audio webhooks includebranch_id,version_idandenvironment. - Conversation history limit: Conversation history, count and search requests from users who can see more than 10,000 agents return
403conversation_history_agent_limit_exceededunless they filter by agent, instead of returning truncated results. - Guardrails: Triggered guardrails record
execution_mode(blockingorstreaming) andguardrail_details, with the judge’s rationale or the classifier verdict, and the conversation transcript shows why a guardrail triggered. Test runs returntriggered_guardrailsand show guardrail triggers in the test transcript. See Guardrails. - Branches: List branches and the branch get and update endpoints return
sync_statusfor branches that update automatically, with theoutcomeandattempted_attime of the last automatic rebase, and the branches table shows its state. Open merge proposals whose target branch is merged and archived are retargeted to the merge target instead of closing. Shareable agent links follow the live traffic split across branches by default, with pinning to the current branch as a separate option. - Defaults and models: Agents created with Create agent have
workflow.prevent_subagent_loopson unless the request sets it. When Eleven v3 Conversational fails during a call, agents fall back to Eleven v4 Turbo, Eleven v4 and then Flash, and never toeleven_v3, so the voice no longer changes mid-call. The fallback ramp forgemini-2.5-flashandgemini-2.5-flash-litenow starts on November 15 instead of November 8; the November 30 cutoff is unchanged. - Realtime API:
response.createdandresponse.donealways includeresponse.status_details.turn_detection.interrupt_response: falseis honored, and only committed agent text is streamed.audio/pcminput at sample rates other than 24000 Hz is rejected withinvalid_audio_format, andaudio/pcmais removed from the audio format enum. Afunction_call_outputthat arrives after its call stopped waiting no longer suppresses the nextresponse.create. See Realtime models. - Knowledge base: Sitemap crawls have an Auto Discover option that adds new sitemap pages at the folder’s auto-sync frequency. Test retrieval in the dashboard covers every retrieval type. The knowledge base tool is registered only when the agent has documents attached. See Knowledge base.
- Channels and telephony: Pre-call webhook
request_headersaccept environment variables, in the API and the dashboard. The Salesforce Cases trigger accepts an optional Reply-To address for email replies. Activating a second Telegram trigger for the same bot returns409instead of replacing the first. Redelivered WhatsApp messages no longer get a second reply. Amazon Connect calls no longer end with408when the agent pauses.transfer_to_agentconditions resolve dynamic variables. WebRTC calls drop stale audio after an interruption, and an inbound audio batch longer than 15 minutes keeps the newest 15 minutes instead of ending the call. - Conversation WebSocket: Text frames that are not a JSON object close the connection with
1008instead of1011. Error closes give end users a generic reason; the detailed reason is sent only to sessions of the agent’s owner. - Privacy: Conversation history redaction runs on each turn of asynchronous channels such as Slack, email, WhatsApp text and Telegram before the turn is stored. With history redaction on and audio recording off,
audiopost-call webhooks and storage exports deliver entity-redacted audio. Triage no longer clusters issues from agents in zero retention mode workspaces. - Dashboard: Agents Home (
/app/agents) opens Architect, and the previous Home dashboard moves to Dashboards (/app/agents/analytics). Architect’s Pro and Lite menu lets you choose a reasoning effort on every plan. Concurrency charts count only voice conversations and no longer change with the chart granularity. Saved views keep every filter, agent search matches partial words in agent names, the Tools tab groups statistics by tool ID, and batch call CSV exports work while calls are still in progress. Triage tickets can be sorted bylast_reported.show_rich_contentcarousels accept up to 10 cards, up from 5. - Evaluations and analysis: Run conversation evaluation accepts
evaluated_agent_idandevaluated_branch_id, so reruns work for conversations transferred to another agent; rerunning requires edit permission. Analysis item references without a pinned version storeversion_idas"latest". Tool latency statistics used forautopre-tool speech exclude MCP approval waits and failed calls. - Fixes: Conversations on Claude 5.5 models no longer fail with
400when a response interleaves thinking with text or tool calls. Test runs from the agent editor compile procedures, so structured procedures no longer fail with a transfer error. Merges and rebases no longer report conflicts when both branches made the same procedure change, or on computed test fields and MCP settings. Duplicating an agent remaps procedure references in structured procedure steps. Conference transfers on outbound SIP trunk calls work when the number has no assigned agent. Workflow nodes with tool overrides that previously failed to save now save, new backward transitions no longer show anexit_sloterror, andenumvalues on LLM expression schemas are kept on save. Calls whose environment does not exist are markedfailedwith the reason inmetadata.error. Architect works again in the India and Singapore data residency environments.
Workspaces
- Credit limits:
PUT /v1/workspace/members/{user_id}/usage-limitsets the monthly credit limit (usage_limit) of a member or service account, or removes it withnull. Admins can also set it from the Credit limit button on the service accounts page. Personal credit limits now count API key and agent usage for every seat type except basic, which still counts only Creative usage. API key and agent usage from before this change is not counted retroactively. See Members. - Billing group concurrency: Enterprise admins can set Text to Speech, Dubbing and Music concurrency limits on billing groups from the group dialogs. See Billing groups.
- Architect permission: Existing full and read-only Agents grants, including grants to Everyone and to groups, now also grant the Architect permission.
- Auth connections: OAuth2 JWT auth connections add
jwt_usage, whereclient_authenticationsends the JWT as aclient_assertionon aclient_credentialsrequest. OAuth2 JWT and private key JWT connections addextra_headersfor additional JWT header parameters, such asx5t, and thePS256,PS384andPS512algorithms. See Create auth connection. - Webhooks: OAuth2 webhooks can send client credentials in a Basic authorization header (
token_endpoint_auth_method: client_secret_basic).GET /v1/workspace/webhooks?include_usages=trueno longer times out for workspaces with many agents, and webhook retries no longer deliver the same job twice. - Audit logs: The
activity_nameandclass_namequery parameters of List audit logs are typed as theAuditActivityNameandAuditClassNameenums, which list every accepted value. - Fixes: SCIM requests that time out waiting on a group return
429withRetry-After: 5instead of500. Saving the Everyone group no longer fails in workspaces with a permission allowlist.
Sound effects
- Concurrency limits: Sound effects have their own per-plan concurrency limits instead of the Text to Speech limit: 4 on Free and Starter, 6 on Creator and Pro, and 8 on Scale and Business. Requests beyond the limit and its reduced-priority queue return
429. Enterprise customers can request a higher limit. The usage analytics concurrency chart shows Sound effects with its limit. See Sound effects.
Speech Engine
- Authentication:
auth.enable_auth(boolean, defaultfalse) on Create Speech Engine and Update Speech Engine requires a signed URL or token to start a conversation. The Security tab has an Enable authentication toggle. - Sharing: Workspace sharing endpoints accept a Speech Engine’s
seng_ID, so Speech Engines can be shared in a workspace.
Hosted MCP
- Agent inspection tools: 24 tools for agent analytics, configuration summaries, insights, alerts, test summaries, branch conflicts and conversation counts, available with the existing
convai_readandconvai_writepermissions. See Hosted MCP. - Prompt editing: MCP clients can find and replace or append text in an agent’s system prompt or first message on a branch. The edit is staged in the branch draft and goes live when the draft is published.
- Paginated tool listing:
tools/listis returned in pages of at most 256 KiB, so MCP clients that reject large listings can connect. - Fixes: OAuth clients that open the authorization page in a popup, such as Gemini Enterprise, can connect. Media in the Flows view loads in ChatGPT.
Speech to Text
- Scribe v3 and Scribe v3 Live: Create transcript accepts
scribe_v3asmodel_id, with newnumbers_format(digitsorwords) andsecondary_language_codesparameters. The realtime WebSocket acceptsscribe_v3_live, withcommit_strategy=turn_prediction,multilang_modeand a keyterm library of up to 10,000 terms. Scribe v2 and Scribe v2 Realtime remain available. See Models. - Timestamps with background filtering: The realtime WebSocket accepts
include_timestampstogether withfilter_background_audio, with timestamps mapped to the original audio. - Lower partial latency: Realtime partial transcripts are sent as soon as they are produced, removing up to 100 ms of delay.
turn_probabilityon partial transcripts is rounded to two decimal places. - EU data residency:
transcript_editis available in the EU data residency environment. - Fixes: Realtime transcripts no longer contain emotion labels such as
{neutral}.
Voices
- Voice Library matches: When Instant Voice Clone samples match a protected voice that is available in the Voice Library, Create IVC voice and
POST /v2/voices/create-preview-from-samplesreturnlibrary_voice_match, withlibrary_voice_idandpublic_owner_id. If the clone is blocked, the403response includes the same object indetail.data. Pass both IDs toPOST /v1/voices/add/{public_user_id}/{voice_id}to add the library voice instead. - Voice categories: The
VOICE_CATEGORYenum no longer listsfamous. - Fixes: Instant Voice Clone drafts are stored per user and workspace, and the page no longer loops on an old draft. Adding a library voice to My Voices no longer reports success without listing the voice.
Dubbing
- Failed languages: Target languages that cannot start, for example because the quota is exceeded, are marked
FAILEDinstead of stayingPENDING, and targets pending for more than 7 days fail. - Short sources: Sources shorter than 0.1 seconds are rejected with
source_too_shortbefore any charge. - Verified clones: Dubbing accepts a workspace’s own verified voice clones that match a protected voice instead of blocking them.
- Legacy manual mode: The Manual mode option is renamed Legacy manual mode and is shown only when Dubbing v1 is selected.
Studio
- Aspect ratios: Projects can use 3:4 and 4:3 aspect ratios, and media in those ratios starts new projects at the matching ratio.
- Subtitle timing: SRT and VTT exports apply paragraph trims, so subtitles match the exported audio.
- Faster blank projects: Blank audiobook projects are created without a wait, and blank
eleven_v3projects no longer get the sample template. - Fixes: Replacing the content of a large project with
POST /v1/studio/projects/{project_id}/contentno longer fails with500. IME input, such as Japanese, is no longer committed early or mid-word in the editor. Pause after paragraph pauses at the real end of a paragraph that is still loading, and no longer pauses as soon as you press Play at the start of a paragraph.
Image and video
- Image editor: Rotating a layer rotates its selection box, masks sit in their own draggable box, and text and shape layers can be rotated, masked and recolored.
- Flows: Figma
.figimports keep small icons whole and keep drop shadows, rounded clipping and borders of stroke-only frames. The Text to Speech node offers the full 0 to 100% Stability range for Eleven v4. Seedance 2.5 unlocks the aspect ratio when a video reference is selected, and dropped assets are arranged in a grid.
Voice Changer and Voice Isolator
- Video in Voice Changer: Voice Changer in the app keeps video uploads as video. History opens a viewer with the original and converted audio tracks, and the converted video can be downloaded as MP4.
- Fixes: Voice Isolator returns
400invalid_audioinstead of500forpcm_s16le_16uploads with an odd number of bytes.
Text to Speech
- Pronunciation previews: The pronunciation dictionaries dialog has a model picker for previewing rules, instead of always previewing with
eleven_turbo_v2.
Billing
- Voice library surcharge: The agents voice library surcharge is charged per voice, so usage analytics attributes it to each voice instead of “No voice”.
Other products
- AI Speech Classifier: The classifier accepts video files and checks their audio track.
SDK releases
JavaScript SDK
- v2.77.0 - Added a
Workflowbuilder, exported from the package root, that builds agent workflows with chained calls and produces theworkflowvalue foragents.create()andagents.update(). See Workflows as code.
Python SDK
- v2.77.0 - Added the
Workflowbuilder inelevenlabs.agent_workflows, which builds agent workflows with chained calls and produces theworkflowvalue foragents.create()andagents.update(). Speech Engine create and update accept theauthparameter.
ElevenLabs CLI
- v1.9.1 -
agents workflow diagramdraws bare workflow files exported by the SDKWorkflowbuilders.onboardwaits up to 30 minutes for sign-in approval instead of 5, retries a token request that could not connect, and reports a copied key that/v1/userrefuses as “couldn’t be checked” instead of belonging to another account.
Unity SDK
- v0.1.0 - First release of the ElevenAgents Unity SDK (
io.elevenlabs.agents), installed from the git URL through the Unity Package Manager. Adds theConversationsession API, conversation events, messaging, client tools, audio and device controls, a native WebSocket transport for the Editor and desktop players, and a WebGL transport with WebSocket and WebRTC. Requires Unity 6.3 LTS or later. See Unity SDK.
Packages
- @elevenlabs/client@1.35.0 and @elevenlabs/types@0.29.0 -
overrides.tts.modelIdis typed asConversationConfigOverrideTtsModelId, which lists the documented ElevenAgents TTS models. Other strings are still accepted.model_idis added toconversation_config_override.ttsonly whenmodelIdis set. - @elevenlabs/convai-widget-core@0.25.0 - The voice and text mode toggle is a Voice/Text segmented control, with
voice_mode_labelandtext_mode_labeltext contents. Adefault-sizeattribute (compactorexpanded) sets the size the chat opens at. A “new messages” button with a count appears when messages arrive while the transcript is scrolled up, labeled with thenew_messageandnew_messagestext contents. - @elevenlabs/convai-widget-core@0.25.1 - Carousels show up to 10 cards, up from 5.
- @elevenlabs/react@1.21.1, @elevenlabs/react-native@1.2.38, @elevenlabs/convai-widget-embed@0.25.0 and 0.25.1 - Updated to the latest
@elevenlabs/clientand widget releases.
API changes
New endpoints
Workspaces
PUT /v1/workspace/members/{user_id}/usage-limit- Set or clear the monthly credit limit of a workspace member or service account. Requiresusage_limit(integer, minimum 0, ornullto remove the limit). Admins only.
Updated endpoints
ElevenAgents
- Create agent and the agent get and update endpoints
- TTS configuration adds
clone_strength(number, default0), supported voices add nullableclone_strength(number), and TTS conversation and workflow overrides add nullableclone_strength. The override config addstts.clone_strength(boolean, defaultfalse). - Prompt overrides add nullable
reasoning_effort(LLMReasoningEffort), and the override config addsprompt.reasoning_effort(boolean, defaultfalse).conversation_config_overrideon the/v1/convai/conversationWebSocket acceptsagent.prompt.reasoning_effortandtts.clone_strength. - Pre-call webhook
request_headersvalues acceptConvAIEnvVarLocatorin addition to strings and secret locators. - Analysis item references (
AttachedUserEvaluationRef,AttachedUserDataCollectionRef) defaultversion_idto"latest"and it is no longer nullable. SearchStrategydefaults tosemantic.DataCollectionValueSchema.allowed_values_dynamic_variableis deprecated; useallowed_valuesinstead.
- TTS configuration adds
- List conversations - Added nullable
cost(integer) andcost_fiat(number) to each conversation summary. - Get conversation details
- Asynchronous conversation metadata adds nullable
integration_trigger_id(string) andplatform_metadata. - Transcript rows add nullable
created_at_unix_ms(integer), also in test responses. - Triggered guardrails add nullable
execution_mode(streamingorblocking) andguardrail_details(LLM, classifier or unknown verdict). AgentDefinitionSourceaddsmcp.
- Asynchronous conversation metadata adds nullable
- Run conversation evaluation - Added optional
evaluated_agent_idandevaluated_branch_id(string). - List branches and the branch get and update endpoints - Added nullable
sync_statuswithlast_attempt(outcome:succeeded,conflict,no_common_ancestor,rejectedorerror, andattempted_atin Unix seconds). - Test runs (
POST /v1/convai/agents/{agent_id}/run-testsand the test invocation endpoints) - Addedtriggered_guardrails(array). - Triage tickets (
GET /v1/convai/triage-ticketsandGET /v1/convai/agents/{agent_id}/triage-tickets) - The sort enum addslast_reported. - Realtime client secrets (
POST /v1/convai/realtime/client_secrets) -AudioFormatno longer listsaudio/pcma.
Speech Engine
- Create Speech Engine and Update Speech Engine - Added optional
authwithenable_auth(boolean, defaultfalse). Speech Engine responses includeauth.
Voices
- Create IVC voice - Added nullable
library_voice_matchwithlibrary_voice_idandpublic_owner_id(string). VOICE_CATEGORY, used by the dubbing similar voices endpoint, removedfamous.
Workspaces
- Create auth connection and the auth connection get and update endpoints
- OAuth2 JWT connections add
jwt_usage(authorization_grant, the default, orclient_authentication). - OAuth2 JWT and private key JWT connections add
extra_headers(map of strings), andalgorithmaddsPS256,PS384andPS512.
- OAuth2 JWT connections add
- List audit logs -
activity_nameandclass_namequery parameters changed from string to theAuditActivityNameandAuditClassNameenums.
SFX v3
SFX v3 (eleven_text_to_sound_v3), our newest sound effects model, is now available to every workspace. Compared with eleven_text_to_sound_v2, it follows prompts that contain several instructions more closely, handles prompts with music or vocals better, sounds more natural and is less likely to clip on high-frequency sounds such as chimes and whistles.
In ElevenCreative, SFX v3 is the default model in Sound Effects, Studio and Flows, and the sound effects that Video to music adds are generated with it. In the API, set model_id to eleven_text_to_sound_v3 on Create sound effect, and set loop to true for a sound effect that loops smoothly. eleven_text_to_sound_v2 is deprecated and will be removed on March 1, 2027.
Eleven Realtime for every workspace
ElevenAgents voice agents in every workspace can now use Eleven Realtime, completing the gradual rollout that started on October 19. Pick Eleven Realtime v1 Mini or v1 Max as the agent’s voice mode, or connect an OpenAI Realtime client to the Realtime WebSocket, and review sessions on the Realtime Conversations page. Realtime conversations can also be organized into groups through the API, under /v1/convai/realtime/conversation-groups.
Professional Voice Clones on Eleven v4
Text to Speech Professional Voice Clone fine-tuning for Eleven v4 and Eleven v4 Turbo is now available to every workspace. To train an existing clone, hover over it in My Voices and click the plus button next to Eleven v4. Requests for a clone fine-tuned on v4 now use the clone_strength saved in the voice’s settings when the request sets neither clone_strength nor use_pvc_as_ivc, so you no longer need to pass it on every call. See Eleven v4.
Analysis items API
ElevenAgents evaluation criteria and data collection items can now be managed through the API as reusable analysis items. Create, list, update, copy and delete them under /v1/convai/analysis-items, attach or detach them across agents and branches with a single request, list each item’s versions, and check which agents depend on an item before you change or delete it. Inline evaluation criteria and data collection in agent configurations keep working and are stored as attached items. See Analysis library.
Noisy environment detection in the API
ElevenAgents noisy environment detection can now be configured through the API and CLI. Set conversation_config.turn.noisy_environment_detection to true and the agent speaks noisy_environment_message once, without interruption, when it hears other people speaking in the caller’s background, asking them to move somewhere quieter. Add a translated message for each additional language in language_presets. The setting is available for voice agents in Enterprise workspaces; see Noisy environment detection.
Realtime entity redaction
Speech to Text can now redact entities in realtime transcripts. Pass entity_redaction and entity_redaction_mode as query parameters on the realtime WebSocket, in the same format as batch transcription, and each committed transcript is followed by a redacted_transcript event carrying the original text and the redacted_text. The Python and JavaScript SDKs from v2.76.0 and @elevenlabs/client from 1.34.0 support these options. See Entity detection.
All changes
ElevenAgents
- Analysis items API: Evaluation criteria and data collection items can be managed as reusable analysis items under
/v1/convai/analysis-items, with endpoints to list, create, get, update, copy and delete items, list versions, list dependent agents and update attachments in bulk. - Noisy environment detection:
turn.noisy_environment_detection(defaultfalse) andturn.noisy_environment_messageare available in the API, along withnoisy_environment_message_translationon language presets andnoisy_environment_messagein conversation and workflow overrides. Available for voice agents in Enterprise workspaces. See Noisy environment detection. - Eleven Realtime available to all workspaces: Realtime voice mode, the Realtime WebSocket and the Realtime Conversations page are available to every workspace, completing the gradual rollout announced on October 19. Realtime conversation groups can now be managed through the API, under
/v1/convai/realtime/conversation-groups. - Realtime API behavior: Each manual
input_audio_buffer.committranscribes only the speech since the previous commit, andinput_audio_buffer.cleardrops buffered audio.response.metadatafromresponse.createis echoed onresponse.createdandresponse.done, with at most 16 string pairs.input_audio_buffer.speech_startedandspeech_stoppedare sent once per committed user turn instead of on every pause. Idle audio sessions close after 3 minutes instead of 60 seconds, after asession_idle_timeouterror. - GPT-6.1 Terra:
gpt-6.1-terracan be selected as the agent LLM and as the analysis LLM, on the standard and fast service tiers, with reasoning effort fromnonetomax(defaultmedium). - Default LLM: New agents created without an explicit LLM use
gemini-3.5-flashinstead ofgemini-2.5-flash, and the default backup LLM cascade leads with Gemini 3.5 Flash models. Existing agents are unchanged. Gemini 3.5 Flash-Lite is also available in the India and Singapore isolated environments. - Model deprecations:
gpt-4.1-nanoandgpt-4o-2024-05-13, which OpenAI shut down on October 23, are deprecated immediately, and agents using them run ongpt-5.6-lunaandgpt-5.6-solrespectively.gpt-5,gpt-5-mini,gpt-5-nanoand their2025-08-07snapshots are deprecated ahead of OpenAI’s December 11 shutdown, with traffic moving togpt-5.6-sol,gpt-5.6-terraandgpt-5.6-lunain steps from 10% on November 11 to 100% on December 6. List LLMs returnsdeprecation_infofor these models. Claude Sonnet 4.5 is deprecated on November 30, with Claude Sonnet 4.6 as the replacement, and is removed from model selectors. The fallback ramp forgemini-2.5-flashandgemini-2.5-flash-litenow starts on November 8 instead of November 1; the November 30 cutoff is unchanged. - Conversation message counts: List conversations and Get conversation details return
user_message_countandagent_message_count.message_counton the conversation list now counts only turns with text, so tool-only turns are no longer counted. Conversations with no messages get acall_successfulvalue ofunknown, and the dashboard shows them as “No conversation”. - Knowledge base: Update document accepts
enable_auto_sync,auto_removeandminimum_frequency_days, and the dashboard can edit these auto-sync settings after a document is created, optionally applying folder settings to existing documents. Sitemap crawls acceptexclude_patterns, with an Exclude patterns field in the sitemap step of the URL dialog. Website crawls deduplicate pages by canonical URL, and Google Drive auto-sync pauses when the Drive token is rejected and resumes after you reconnect. See Knowledge base. - Agent testing: Tool call tests accept
guardrail_triggerto assert that a guardrail should or should not trigger on the generated turn, with Should trigger and Should not trigger options in the dashboard. Tests whose turn is blocked by a guardrail now fail with “Agent response blocked by guardrail” instead of returning an inconclusive result. List test invocations acceptscreated_by_user_id, and the test history adds a creator filter. Test runs are returned in creation order. See Agent testing. - Branches and merges: Branches can have a default environment, set when creating or publishing to a branch and changed from the branch menu. Conversations on the branch use that environment unless the caller requests another one, and such branches cannot receive live traffic. Merge and rebase conflicts can be resolved with Architect, and procedure conflicts can be edited and combined like prompt conflicts. A single procedure’s unpublished changes can be discarded without discarding the whole draft. Branches whose only version was created with changes can be merged, and branch responses report a
current_live_percentageof 100 for the main branch when the agent has no live deployment. - Workflows: Update state nodes and the
update_statetool can override LLM settings (model, temperature, token limit and reasoning) for the rest of the conversation from a new LLM tab. Double-click a workflow edge to add reroute points that shape it. Workflow tool nodes accept only configurable system tools plusstart_procedureandend_procedure. - Issues: Issue descriptions can be edited from the issue details sheet, and Update triage ticket accepts
qa_comment. See Triage. - WhatsApp: WhatsApp conversations record
call_result(answered,voicemail,busy,no_answerorfailed), and conversations opened with a template recordopening_messagewith the template name, delivery timestamps, error code and pricing category. Call permission requests recordresponded_at_unix,permanentandgranted_by_user_call. WhatsApp batch calls that are declined, missed or cannot be placed mark the recipientfailedinstead of leaving itinitiated. See WhatsApp. - Widget: Widget configuration adds
transcript_speech_sync_enabled(defaulttrue), with a Reveal transcript as the agent speaks switch; set it tofalseto show each message in full at once.show_rich_contentcarousel cards acceptrating(0 to 5) andreview_count, shown as stars. Resumed persistent text conversations replay rich content in their history. See Widget. - Channel integrations: Asynchronous channels such as Slack, Telegram, Intercom, Gmail, Salesforce and Custom Channel record message timestamps from the source platform and store the conversation’s
main_languagefrom the language override or the agent’s language instead of alwaysen. Salesforce passes the customer’s email to the agent asintegration__salesforce_customer_email, which is not a verified identity. Gmail triggers run on the agent’s current version, back off on Gmail rate limits and set aside threads that keep failing. - Telephony: Twilio conference transfers can combine post-dial digits with call screening (
require_acceptance). Tools that wait for speech to finish, such astransfer_to_numberandend_call, are skipped if the caller hangs up during their pre-tool speech. Callingskip_turnafter a turn timeout starts the skip-turn wait, so the silence timeout no longer ends the call during it. - Environment variables: Deleting an environment variable that the tip of a live branch references fails with
409and lists the affected agents and branches. The environment variables table adds Remove value and Delete row actions. - Dashboard: The conversation view shows human agent replies captured after an escalation in a separate post-escalation section. The system prompt editor’s wand opens an Ask Architect menu to review and improve the prompt or write a new one. Filtered conversation exports include evaluation criteria and data collection results. Turns where the agent speaks after a silence timeout no longer report end-to-end latency. The agent voice settings show the Clone strength slider for every voice fine-tuned for a v4 model.
- Fixes: Files that the LLM provider cannot process are dropped instead of failing later turns. Deleting an agent removes its trigger connections, so the integration connection can be deleted afterwards. The prompt editor saves
&and>as typed instead of encoding them. Say nodes no longer duplicate messages, merge proposals that add a node group no longer fail, and node groups show their configured titles. The structured procedure editor opens procedures authored through the API, and dynamic variables in If step expressions are included inreferenced_dynamic_variables. Agents whose data collection item uses a dynamic variable can be published again. Knowledge base rediscovery adds pages to older crawl jobs. The concurrency chart no longer overstates peaks when grouped by agent or branch. Garbled characters in LLM tool call arguments are repaired, and WhatsApp webhooks with resent interactive replies or contacts without a name are accepted instead of failing with422.
Speech to Text
- Realtime entity redaction: The realtime WebSocket accepts
entity_redaction(all, a category such aspii,phiorpci, or a list of entity types) andentity_redaction_mode(redacted,entity_typeorenumerated_entity_type, the default). Each committed transcript is followed by aredacted_transcriptevent, and the settings are echoed insession_started.entity_redactioncannot be combined withtranscript_edit. See Entity detection. - Realtime authentication: Bearer tokens passed to the realtime WebSocket in
Sec-WebSocket-Protocolrequire the Speech to Text permission, as the other authentication methods do. - Usage on empty responses: Empty and silent responses from Create transcript include
audio_duration_secs. - Fixes: Webhook transcription requests, with a file upload or
cloud_storage_url, now applydiarization_thresholdandseed.
Sound effects
- SFX v3 for everyone: SFX v3 (
eleven_text_to_sound_v3) is available to every workspace and is the default model in Sound Effects, Studio and Flows. The sound effects that Video to music adds withadd_sfxare generated with SFX v3. See SFX v3.
Text to Speech
- Professional Voice Clones on Eleven v4: Professional Voice Clone fine-tuning for Eleven v4 and Eleven v4 Turbo is available to every workspace. See Eleven v4.
- Saved clone strength: Eleven v4 requests for a fine-tuned Professional Voice Clone use the voice’s saved
clone_strengthwhen the request sets neitherclone_strengthnoruse_pvc_as_ivc. - Pronunciation dictionary limit: Applying pronunciation dictionaries can expand the input text to at most 50 times its length plus 10,000 characters. Requests over the limit return
400pronunciation_dictionary_expansion_too_large. - Fixes: History items for v4 generations keep
clone_strengthandspeedin their saved settings.
Studio
- Eleven v4 recommended: Project settings and the model picker recommend Eleven v4 instead of Multilingual v2.
- Larger media uploads: Pro and higher plans can upload audio and video assets of up to 3 GB, up from 1 GB.
- Word uploads: Audiobook projects can be created from Word
.docxfiles as well as EPUB and PDF. - Video clip background removal: Video clips have a Remove background action, and right-clicking a clip opens its AI Tools.
- Pronunciations tab: Audiobook projects have a pronunciations sidebar tab, opened from the selection menu or with Mod+Shift+P.
- Fixes: The import speech endpoints for external audio and video clips return
429system_busyand refund the transcription charge when transcription capacity is unavailable, instead of returning500. Duplicating a project no longer fails when an image clip points at a source image that is not copied.
Realtime Translation
- Close codes: The WebSocket always closes with a code:
1000for a normal end,1008for a rejected request such as an unknownvoice_idor an authentication failure, and1011for a server error. - Invalid frames: Frames that are not a JSON object receive an
input_errormessage and the session continues. - Fixes: Translated audio is encoded in the requested
output_format, so μ-law and A-law output plays correctly. Repeated utterances are no longer dropped, Chinese, Japanese and Korean clauses are no longer spoken twice or skipped, and Korean word order is improved.
Image and video
- SVG assets: SVG uploads to Assets stay SVG and render as vectors, including in the image editor. Scripts, event handlers and remote references are removed, models receive a PNG, and asset responses omit
preview_content_urlfor SVGs. - Flows: Node groups and the structure sidebar are available in every workspace. LLM nodes on video-capable models such as
gemini-3.5-flash-llmaccept video prompt references. - Clearer reference errors: Generations whose reference file was refused on upload, or is not decodable audio, fail as a validation error with a clear message instead of an unexpected error.
Speech Engine
- Longer conversations: Speech Engine conversations can last up to 2 hours, up from 40 minutes. See Speech Engine.
- Request headers: Speech Engine request header values can reference workspace secrets and are validated in the dashboard. Existing headers must be cleared before the WebSocket URL can change.
Voices
- Larger voice selections: Manual voice selection is no longer limited to 10 voices, and link-shared voice collections offer Select all. Shared collection links open in the EU, India and Singapore data residency environments.
- Fixes: The voice verification captcha accepts Serbian readings in Latin or mixed script.
Music
- Higher upload limit: Upload music accepts up to 100 songs per user per 24 hours, up from 20.
Workspaces
- Architect permission: Architect access is controlled by its own workspace permission.
PermissionTypeaddsagents_architectand group permissions addarchitect, so admins can turn off Architect without removing Agents access, and API keys have a separate Architect scope. - History download permission: Download history items requires the
speech_history_readAPI key permission. - Fixes: New users whose only invite is to a child workspace that inherits or borrows SSO can sign up with that SSO instead of receiving
400provider_not_allowed.
Billing
- Architect charging: Architect usage is now billed from workspace credits as its own line item, separate from agents’ minutes and LLM cost. Billing is rolling out gradually; the composer in the dashboard shows when it applies to your workspace. See Pricing and usage.
- Realtime pricing: Workspace pricing tables list Realtime API audio (
eleven_realtime_v1_audio) and the Eleven Realtime v1 Mini and Max token rates. - Cached LLM tokens: Cache writes are billed at least at the input rate and cache reads at the vendor’s cached rate. LLM calls for post-call analysis, guardrails and agent tests bill cached prompt tokens at those rates instead of the full input rate. Gemini 2.5 Flash-Lite cache reads are billed at 0.01 instead of 0.
- Preview conversation discount: The discount on conversations started from the agent tester applies only to workspaces on 2022 pricing. On 2026 pricing,
charging.dev_discountisfalse.
Hosted MCP
- Fixes: OAuth discovery names the host the client connected to, so MCP clients such as Claude Code, Cursor and VS Code can connect to
https://api.elevenlabs.io/v1/mcp.
SDK releases
JavaScript SDK
- v2.76.0 - The realtime Speech to Text wrapper accepts
entityRedaction. Addedclient.conversationalAi.analysisItemsfor managing analysis items, and methods onclient.conversationalAi.realtimeto list, create, get, update and delete realtime conversation groups and to set or clear a conversation’s group.
Python SDK
- v2.76.0 - The realtime Speech to Text wrapper accepts
entity_redaction. Addedclient.conversational_ai.analysis_itemsfor managing analysis items, andlist_conversation_groups,create_conversation_group,get_conversation_group,update_conversation_group,delete_conversation_group,set_conversation_groupandclear_conversation_grouponclient.conversational_ai.realtime.
Android SDK
- v0.13.1 -
textOnlyis now sent to the server, so text-only sessions run as text conversations instead of voice conversations. The top-leveltextOnlysetting takes precedence overoverrides.conversation.textOnly, as in the JavaScript SDK, and setting only the override now selects the WebSocket transport.
Unity SDK
- v0.1.0 - First release of the Unity SDK for voice agents, installed through the Unity Package Manager from a git URL. It requires Unity 6.3 LTS or later and runs in the Editor, desktop players and WebGL, with WebRTC available on WebGL.
Conversation.StartSessionAsyncstarts a session with an agent ID, signed URL or conversation token, and the SDK supports client tools, conversation events, messaging, feedback, device selection and spatial audio through anAudioSource. GettingStarted and QuickStart samples are included.
ElevenLabs CLI
- v1.9.0 -
agents workflow diagramdraws an agent workflow in the terminal, or prints it as Mermaid with--format mermaidor--format mermaid-md.agents graphshows how agents, tools, tests, knowledge base documents and handoffs connect across the agents inagents.json, and--useslists the agents that reference a given tool or document. Both commands read local files only. Regenerated against the latest API. - v1.8.0 - Request
User-Agentheaders include how the CLI was installed (install/<source>, such as npx, Homebrew or the installer). Pass--telemetry off, or setELEVENLABS_TELEMETRY=offorDO_NOT_TRACK=1, to drop these tags. Regenerated against the latest API.
Packages
- @elevenlabs/client@1.34.0, @elevenlabs/types@0.28.0 and @elevenlabs/react@1.21.0 -
Scribe.connect()anduseScribe()acceptentityRedactionandentityRedactionMode, and dispatch theredacted_transcriptmessage after each committed transcript.useScribe()adds anonRedactedTranscriptcallback and setsredactedTexton the matching entry incommittedTranscripts.stripAudioTagsremoves audio tags written in any language, not only tags in English letters. - @elevenlabs/convai-widget-core@0.24.0 - With
persistent-sessionenabled, a voice call can be resumed after a page navigation. The next page shows the conversation so far with “Resume talking?”, and resuming starts a new conversation that continues from those turns, skipping the greeting if the agent allows first message overrides. - @elevenlabs/client@1.32.0 - Self-hosted worklets (
workletPaths) are checked against the installed client version. A worklet copied from another release fails at session start with an error that names the worklet and path. Re-copy the files from@elevenlabs/client/worklets/*when upgrading. - @elevenlabs/client@1.32.1 - Fixes the start of the agent’s first message playing too fast and too high when the browser cannot use the agent’s sample rate. Voice sessions wait for the output resampler and fail with an error if it cannot be created.
- @elevenlabs/client@1.33.0, @elevenlabs/types@0.27.0 and @elevenlabs/convai-widget-core@0.23.0 - Added a
cloneStrengthTTS override to the client and anoverride-clone-strengthattribute to the widget. History rows of resumed persistent text conversations can carryrich_content, and the widget replays it. - @elevenlabs/react@1.20.3 -
useScribewithautoConnectkeeps a single open session under React StrictMode. - @elevenlabs/convai-widget-core@0.22.0 - Added the
transcript-speech-syncattribute; set it tofalseto show the agent’s full message at once in voice calls. Carousel cards accept an optionalratingfrom 0 to 5 andreview_count. - @elevenlabs/convai-widget-core@0.21.6 and 0.22.2 - Voice calls show typing dots when the agent takes more than a second to start replying, or “Thinking…” under the avatar when the transcript is off. Partially revealed markdown links show their text instead of
[blocked], carousel arrows no longer flicker, and carousel images use a 4:5 ratio. - @elevenlabs/react@1.20.1, 1.20.2, @elevenlabs/react-native@1.2.34, 1.2.35, 1.2.36, 1.2.37, @elevenlabs/convai-widget-core@0.22.1, 0.24.1 and @elevenlabs/convai-widget-embed@0.21.6 to 0.24.1 - Updated to the latest
@elevenlabs/clientand widget releases.
API changes
New endpoints
ElevenAgents
- Analysis items
GET /v1/convai/analysis-items- List analysis items, with optionalsearch,kind(evaluationordata_collection),analysis_item_ids,created_by_user_id,sort_direction,page_sizeandcursor.POST /v1/convai/analysis-items- Create an analysis item.GET /v1/convai/analysis-items/{analysis_item_id}- Get an analysis item.PATCH /v1/convai/analysis-items/{analysis_item_id}- Update an analysis item.DELETE /v1/convai/analysis-items/{analysis_item_id}- Delete an analysis item. Passforce=trueto delete it even when agents use it, removing it from the dependent agents and branches.POST /v1/convai/analysis-items/{analysis_item_id}/copy- Copy an analysis item.GET /v1/convai/analysis-items/{analysis_item_id}/versions- List an analysis item’s versions.GET /v1/convai/analysis-items/{analysis_item_id}/dependent-agents- List the agents that use an analysis item.PATCH /v1/convai/analysis-items/{analysis_item_id}/attachments- Attach or detach an analysis item across agents and branches.
- Realtime conversation groups
GET /v1/convai/realtime/conversation-groupsandPOST /v1/convai/realtime/conversation-groups- List and create conversation groups. Create requirestitle(string, unique per workspace, up to 64 characters) and accepts optionaldescription(string, up to 500 characters).GET,PATCHandDELETE /v1/convai/realtime/conversation-groups/{conversation_group_id}- Get, update and delete a conversation group.PUT /v1/convai/realtime/conversations/{conversation_id}/conversation-group- Place a conversation in a group, with requiredconversation_group_id. A conversation belongs to at most one group.DELETE /v1/convai/realtime/conversations/{conversation_id}/conversation-group- Remove a conversation from its group.
Updated endpoints
ElevenAgents
- Create agent and the agent get and update endpoints
- The prompt
llmdefault changed fromgemini-2.5-flashtogemini-3.5-flash. TurnConfigaddsnoisy_environment_detection(boolean, defaultfalse) andnoisy_environment_message(string). Language presets add nullablenoisy_environment_message_translation, turn overrides and workflow overrides add nullablenoisy_environment_message, and the override config addsnoisy_environment_message(boolean, defaultfalse).- Widget configuration adds
transcript_speech_sync_enabled(boolean, defaulttrue).
- The prompt
LLMenum addsgpt-6.1-terra, including inconversation_config_overrideon the/v1/convai/conversationWebSocket.- List conversations and Get conversation details - Added nullable
user_message_countandagent_message_count(integer). - Get conversation details
- WhatsApp conversation info adds nullable
call_result(answered,voicemail,busy,no_answerorfailed) andopening_message(template_name,wamid,sent_at_unix,delivered_at_unix,read_at_unix,replied_at_unix,failed_at_unix,error_code,pricing_categoryandbillable). - WhatsApp call permission requests add nullable
responded_at_unix(integer),permanent(boolean) andgranted_by_user_call(boolean). - Agent definition source enum adds
architect.
- WhatsApp conversation info adds nullable
- Update document - Added optional
enable_auto_sync(boolean),auto_remove(boolean),minimum_frequency_days(integer) andtighten_only(boolean, defaultfalse). - Update triage ticket - Added optional
qa_comment(string). - List test invocations - Added optional
created_by_user_idquery parameter (string). - Create test and the other test endpoints - Tool call evaluations add nullable
guardrail_trigger(object withtypeany). - Rebase a branch onto main (
POST /v1/convai/agents/{agent_id}/branches/{branch_id}/rebase) - The request body acceptsapplication/json.
Speech to Text
- Realtime (
/v1/speech-to-text/realtime)- Added optional
entity_redaction(string or array of strings) andentity_redaction_mode(redacted,entity_typeorenumerated_entity_type, defaultenumerated_entity_type) query parameters. - Added the
redacted_transcriptserver message. session_startedconfig addsentity_redaction(array of strings) andentity_redaction_mode(string).
- Added optional
Workspaces
- Service account and workspace group permissions -
PermissionTypeaddsagents_architect, and workspace group permissions addarchitect.
Fast LLM service tier
ElevenAgents can now run supported LLMs on a fast service tier, which trades a per-token price premium for higher token throughput and lower latency. Set service_tier to fast in the agent’s prompt configuration with Update agent, and check which tiers a model supports in the available_service_tiers field returned by List LLMs. The default remains standard, and the LLM usage calculator accepts service_tier to estimate the cost difference.
Knowledge base crawl exclusions
ElevenAgents knowledge base website crawls can now skip parts of a site. Pass up to 50 exclude_patterns to Create crawl job, or fill in the Exclude patterns field in the whole-website crawl dialog, and URLs that match any pattern are skipped even when they match the crawl pattern. Crawl job responses return the patterns that were used. See Knowledge base for crawl setup.
Clone strength for v4 voices
Text to Speech and Text to Dialogue requests now accept clone_strength, a value from 0 to 1 that controls how closely a Professional Voice Clone fine-tuned for a v4 model follows its training samples. Set it in voice_settings on Create speech or in settings on Create dialogue. The recommended value for most use cases is 0. On the Text to Dialogue endpoints, clone_strength now decides whether a v4 PVC runs as a PVC or an IVC, and use_pvc_as_ivc is deprecated.
Expiring service account API keys
Workspaces service account API keys can now expire. Pass expires_at_unix when you create or update a key, or pick an Expire after value in the dashboard, and the key stops working after that time. Expiries must be between 15 minutes and 90 days in the future; omit the field for a key that never expires. See Service accounts.
All changes
ElevenAgents
- Fast LLM service tier: Agents and workflow overrides accept
service_tier(standardorfast) on supported models. List LLMs returnsavailable_service_tiersfor each model, and the LLM usage calculator acceptsservice_tier. - Knowledge base crawl exclusions: Create crawl job accepts
exclude_patterns, and the whole-website crawl dialog adds an Exclude patterns field. Matching URLs are skipped. - Procedures analytics: The Procedures tab on the agent analytics dashboard is available to every workspace. Its Procedure invocations chart shows how often each procedure runs over time, grouped by procedure or by agent, and its tooltip shows whether a procedure is structured or free-form.
- Start procedures at conversation start: Conversation initiation client data accepts
start_procedure_ids, up to five free-form procedures that start when the conversation starts. The agent must haveenable_procedure_ids_from_clientenabled; structured procedures and unknown IDs fail the conversation start. - Structured procedure steps: Structured procedures can use the
end_proceduresystem tool as a step, and the End step is relabeled End conversation. Sub-procedure and end steps now compile to system tool nodes, so agents start and end the intended procedure more reliably. - Channel integrations out of alpha: The Slack, Telegram, Custom Channel and Intercom integrations are no longer in alpha. Freshdesk remains in alpha.
- Slack thread replies: The Slack Channel Message trigger adds a Respond to setting (
response_mode), so an agent can keep replying in a thread after a single mention. Mentioning the agent with onlystop,muteorquietsilences it in that thread until it is mentioned again, and events that Slack delivers twice get one reply. - Test run names: Run tests accepts an optional
name, which test invocations and their summaries return.PATCH /v1/convai/test-invocations/{test_invocation_id}renames a run, and settingnametonullrestores the generated title. - Branch descriptions: Branch descriptions can be edited from the branches table with Update branch (
description, up to 4,096 characters) and appear in the version tooltip and version graph. - Agent duplication keeps traffic splits: Duplicating an agent copies its live traffic split across the copied branches. The resulting deployment records
duplicateas its deployment source. - Delete environment variables:
DELETE /v1/convai/environment-variables/{env_var_id}deletes an environment variable. - Widget fullscreen on mobile: Widget configuration adds
mobile_fullscreen(defaulttrue), with a Fullscreen on mobile switch in the dashboard. Set it tofalseto keep the compact layout on phones. See Widget. - Issues: Triage is renamed Issues in the agents dashboard, and triage tickets are now called issues. Every QA comment and
flag_issue_for_reviewcall opens its own issue, and an issue can be reopened while the conversation has another open issue. The API endpoints keep theirtriage-ticketspaths. See Triage. - Prompt caching guidance: The system prompt editor flags system variables that change on every turn and prevent prompt caching, and Spotlight adds an Improve prompt caching recommendation for affected agents.
- Realtime voice mode: Switching an agent to Realtime voice mode in the dashboard now selects Eleven Realtime v1 Max. Realtime agents always run tool calls in parallel, so the Parallel tool calling toggle is removed from Realtime model settings. See Realtime models.
- Analytics: Workflow analytics can be filtered by channel, and the Guardrails tab is shown to all users.
- Analysis LLMs:
claude-sonnet-5-5can be selected as the analysis LLM for evaluation criteria and data collection, both for the agent and per item. Agents on Claude Sonnet 5.5 can also set reasoning effort tonone. - Conversation cost: Conversation charging adds
voice_surcharge_charge, the credits charged as the voice actor reward for the Voice Library voices a conversation used. It is not included incost. - Call duration and billing: Call charges and stored call duration stop when the call ends, so teardown time after the call is no longer billed or counted. Calls no longer continue for up to 50 seconds after ending when a knowledge base lookup is in progress.
- Telephony: SIP trunk calls forward the
User-To-User(UUI) header to the conversation initiation webhook. API keys with an IP allowlist now authenticate on the Genesys, Avaya, AudioCodes and Amazon Connect connector WebSockets. Batch call recipients whose call cannot be placed are markedfailedinstead of stayingdispatched. - Observability: OpenTelemetry post-call webhook agent turn spans add
elevenlabs.rag.chunksandelevenlabs.rag.used_chunk_ids, matching the JSON post-call webhook. See OpenTelemetry traces. - Guardrails: In voice conversations, the prompt injection guardrail now evaluates the user turn being answered instead of the agent’s previous message. Conversations on asynchronous channels such as Zendesk, Intercom and Custom Channel that a guardrail blocks now report the guardrail as the termination reason.
- Zero Retention Mode: Synthetic voice detection is skipped for conversations in Zero Retention Mode.
- WhatsApp: WhatsApp conversations resumed by a reply to a batch call template keep their
batch_callmetadata and batch ID. - Phone number secrets: Workspace secrets used as a Twilio auth token or an Exotel API token on a phone number show as in use and cannot be deleted while the number depends on them.
- Fixes: Text conversations end as soon as the
end_calltool runs instead of waiting for the next user message. Voice agents no longer go silent when a turn is cancelled before any audio is sent. The soft timeout no longer fires right after the first message or after tool calls on Eleven v4 voices. Realtime agents with sub-agent, agent tool or prompt say nodes that inherit the LLM can be saved and published. Agent tools that use OAuth2 connections no longer fail with expired tokens when the provider rotates refresh tokens. Thenemotron-35-super-vl-nvfp4context limit is raised from 64,000 to 131,072 tokens.
Text to Speech and Text to Dialogue
- Clone strength:
clone_strength(0 to 1) in Text to Speechvoice_settingsand Text to Dialoguesettingscontrols how closely a Professional Voice Clone fine-tuned for a v4 model follows its samples. The recommended value for most use cases is 0. Edit voice settings storesclone_strengthand Get voice settings returns it. - PVC and IVC on v4 dialogue: On the Text to Dialogue endpoints,
clone_strengthnow decides whether a v4 PVC runs as a PVC or an IVC, anduse_pvc_as_ivcis deprecated.
Voice Changer
- New endpoint paths: Voice Changer is documented at
POST /v1/voice-changer/{voice_id}andPOST /v1/voice-changer/{voice_id}/stream. The/v1/speech-to-speech/{voice_id}paths keep working but are deprecated and removed from the API reference. The Python and JavaScript SDKs call the new paths from v2.75.0, and the SDK method names are unchanged for now. In the next major SDK version,client.speechToSpeechwill be deprecated in favor ofclient.voiceChangerto match the endpoint. See Voice changer.
Workspaces
- Expiring service account API keys: Service account API keys accept
expires_at_unixon create and update, between 15 minutes and 90 days in the future. Later expiries are rejected with400invalid_api_key_expiry. On update, pass"clear"to remove an expiry. - Group concurrency limits enforced: The workspace group concurrency limits (
group_concurrency_limit,group_dubbing_concurrency_limitandgroup_music_concurrency_limit) are enforced for Text to Speech, Voice Changer, Music and Dubbing requests. - Voice cloning permissions: Members without the Create instant voice clone or Create professional voice clone permission can no longer create clones, even with general voice access.
- SCIM group seat types: Workspaces that use SCIM without SCIM routing can assign a seat type to SCIM-synced groups from the Groups page.
- Fixes: Delete invite returns
404invite_not_foundfor an invite that does not exist, instead of a500error.
Speech to Text
- Shorter realtime silence thresholds: Realtime Speech to Text honors
vad_silence_threshold_secsvalues below 0.5 seconds, which were previously treated as 0.5. The minimum accepted value is lowered from 0.3 to about 0.19 seconds. - Webhook retries: Transcription and error webhooks are retried on failure when the webhook’s retry setting is on. See Speech to Text webhooks.
Dubbing
- Legacy endpoints deprecated: The
/v1/dubbingendpoints outside/v1/dubbing/project, including Dub a video or an audio file and List dubs, are marked deprecated in the API reference. - Transcript downloads: The dubbing project menu can download target language transcripts as JSON, and as WebVTT subtitles for recent dubs.
Studio
- Media folders: Workspace library files can be organized into folders in the Studio media browser.
- Video clip actions: Video clips offer voice changer and voice isolator actions.
- Video to music: Studio’s video to music workflow uses Music v2.5 by default.
Image and video
- Retired models: Image & Video and Flows requests for a retired model fail before generation with a
410model_retirederror that names the replacement model. Flows preflight reports the error on the affected node. - Fixes: Generations wait for a just-uploaded reference asset to finish processing instead of failing.
Voices
- Fixes: Cantonese Professional Voice Clone samples are no longer rejected when language detection reports Chinese, and adding PVC samples no longer fails with a
500error when language detection is unavailable.
Billing
- Minimum charge: Text to Speech and Speech to Text requests priced below one credit are charged one credit, and refunds return exactly the credits that were charged.
- Business plan image and video pricing: Business plans on the current pricing pay the same image and video credit rate as other self-serve plans, about 17% fewer credits per image and per video second than before.
Realtime Translation
- OAuth scope: OAuth clients can request the
realtime_translationscope.
SDK releases
JavaScript SDK
- v2.75.0 -
client.speechToSpeech.convertandclient.speechToSpeech.streamnow callPOST /v1/voice-changer/{voice_id}and its/streampath. Addedclient.environmentVariables.delete,client.conversationalAi.tests.invocations.updatefor renaming a test invocation, and an optionalnameonclient.conversationalAi.agents.runTests. Also addsserviceTieron the LLM usage calculators,excludePatternson knowledge base crawl jobs,expiresAtUnixon service account API keys,cloneStrengthin voice settings,startProcedureIdsin conversation initiation client data and the widgetmobileFullscreensetting. The legacy dubbinglist,create,getanddeletemethods are marked deprecated, and the branch rebase request typeBranchesRebaseRequestis renamed toBodyRebaseABranchOntoMainV1ConvaiAgentsAgentIdBranchesBranchIdRebasePost.
Python SDK
- v2.75.0 - Added the
callback_disconnectoption toConversationandAsyncConversation. It receives aDisconnectionDetailswith areasonof"user","agent"or"error", plus the WebSocket close code and reason when available, and replaces the deprecatedcallback_end_session. The regenerated client movesclient.speech_to_speechto the/v1/voice-changerpaths and addsclient.environment_variables.delete,client.conversational_ai.tests.invocations.updateand the same new parameters as JavaScript SDK v2.75.0.
Android SDK
- v0.13.0 -
agent_tool_requestis now handled as a notification event instead of a client tool call, so client tools no longer run twice, the first time without parameters. The SDK also no longer crashes when a client tool answers after the connection has closed.
ElevenLabs CLI
- v1.7.1 -
sayuseseleven_v4and new agent templates useeleven_v4_turboby default. Fixesonboardsign-in on Windows and when the token cannot be saved to the keychain.
Packages
- @elevenlabs/client@1.29.0 and @elevenlabs/react@1.19.0 - Added the
onAudioAlignmentPlaybackcallback, called with{ event_id, alignment }when an aligned audio chunk starts playing, for syncing text with speech. It is not called on WebRTC connections. - @elevenlabs/client@1.30.0 - Added the opt-in
stripAudioTagssession option, which removes audio tags such as[laughs]from agent messages passed toonMessageandonAgentResponseCorrection, and exports thestripAudioTagshelper. Stripped text no longer lines up withonAudioAlignmenttimings. - @elevenlabs/client@1.31.0, @elevenlabs/types@0.26.0 and @elevenlabs/react@1.20.0 - Added
overrides.tts.modelIdfor a per-conversation TTS model override. Realtime Speech to Text emitsRealtimeEvents.AUDIO_STARTEDonce the first microphone audio frame is sent, anduseScribein microphone mode reportsconnectedonly from that point. - @elevenlabs/convai-widget-core@0.21.0 and @elevenlabs/convai-widget-embed@0.21.0 - Voice call transcripts are revealed in step with the agent’s speech, and an interrupted reply keeps only what was said. Resuming a persistent text conversation no longer asks the user to accept the terms again.
- @elevenlabs/convai-widget-core@0.21.1, 0.21.3 and 0.21.4 - Text chat shows typing dots while the agent prepares its reply. Rich content carousels show a page indicator, stay clickable after the conversation moves on, and appear in voice calls while the agent is still speaking. On screens narrower than 768px the chat opens fullscreen; add
mobile-fullscreen="false"to the embed to keep the compact layout. - @elevenlabs/react-native@1.2.31, 1.2.32 and 1.2.33 - Updated to the latest
@elevenlabs/clientand@elevenlabs/reactreleases.
API changes
New endpoints
ElevenAgents
DELETE /v1/convai/environment-variables/{env_var_id}- Delete an environment variable. Returns204.PATCH /v1/convai/test-invocations/{test_invocation_id}- Rename a test invocation with requiredname(string, 1 to 256 characters, ornullto restore the generated title).
Voice Changer
- Voice changer -
POST /v1/voice-changer/{voice_id}and Voice changer stream -POST /v1/voice-changer/{voice_id}/stream- The documented paths for Voice Changer.POST /v1/speech-to-speech/{voice_id}andPOST /v1/speech-to-speech/{voice_id}/streamare removed from the API reference and deprecated.
Updated endpoints
ElevenAgents
- Create agent, Get agent and Update agent
- Prompt configuration adds
service_tier(standardorfast, defaultstandard). Workflow node prompt overrides add nullableservice_tier. - Widget configuration adds
mobile_fullscreen(boolean, defaulttrue).
- Prompt configuration adds
- List LLMs - Each model adds required
available_service_tiers(array ofstandardorfast). - Calculate LLM usage and Calculate expected LLM usage for an agent - Added optional
service_tier. - Get widget - Widget configuration adds
mobile_fullscreen(boolean, defaulttrue). - Create crawl job - Added optional
exclude_patterns(array of strings, up to 50 items of up to 2,048 characters each). - Get crawl job and List crawl jobs - Crawl jobs add
exclude_patterns(array of strings). - Knowledge base document responses -
auto_sync_inforemoves the deprecatedauto_discoverfield. - Run tests - Added optional
name(string, 1 to 256 characters). Get test invocation and List test invocations return nullablename. - Update branch - Added optional
description(string, up to 4,096 characters). Pass an empty string to clear it. - Get conversation details
- Conversation charging adds nullable
voice_surcharge_charge(integer). - Agent definition source enum adds
architect.
- Conversation charging adds nullable
- Agent deployments (
GET /v1/convai/agents/{agent_id}/deployments) - Deployment source enum addsduplicate. - Conversation initiation client data, including the
conversation_initiation_client_dataWebSocket event and Submit batch call - Added optionalstart_procedure_ids(array of strings). - Tool type enum removes
api_integration_mcp.
Text to Speech and Text to Dialogue
- Create dialogue, Stream dialogue and their timestamp variants -
use_pvc_as_ivcis deprecated. - Text to Speech
voice_settingsand Text to Dialoguesettings- Added nullableclone_strength(number, 0 to 1). Get voice settings, Get voice and List voices return it.
Dubbing
- List dubs, Dub a video or an audio file,
GET /v1/dubbing/{dubbing_id},DELETE /v1/dubbing/{dubbing_id},GET /v1/dubbing/{dubbing_id}/audio/{language_code}andGET /v1/dubbing/{dubbing_id}/transcripts/{language_code}/format/{format_type}- Marked deprecated.
Workspaces
- Create service account API key - Added optional
expires_at_unix(integer ornull). - Update service account API key - Added optional
expires_at_unix(integer, or"clear"to remove the expiry; omit to leave it unchanged). - List service accounts and List service account API keys - API keys add nullable
expires_at_unix(integer).
Eleven Realtime
Eleven Realtime is a new family of models for ElevenAgents voice agents, rolling out gradually across workspaces. Each one is a complete voice agent configuration. You pick one model, and ElevenLabs handles the LLM, Speech to Text and Text to Speech settings behind it. Your agent’s system prompt, voice, tools and knowledge base work the same as before.
There are two to choose from. Eleven Realtime v1 Mini (eleven_realtime_v1_mini) is the default, and it’s the fastest and cheapest. Eleven Realtime v1 Max (eleven_realtime_v1_max) reasons before it answers, so use it when your agent has complex instructions or multi-step tool calls.
If you’ve already built on OpenAI’s Realtime API, you don’t need to rewrite your client. The events and the WebSocket transport are the same, so you change the endpoint and the session configuration. Your backend calls Create realtime client secret to get a short-lived secret, and the browser uses it to connect directly.
Realtime Translation
Realtime Translation turns speech in one language into speech in another while the speaker is still talking. It’s a WebSocket API, available now for enterprise workspaces.
You open a connection with a target_language and a voice_id, then stream audio in. You get translated speech back in that voice. You don’t have to say which language the speaker is using, because it’s detected automatically, but you can set source_language if you know it. Along the way, the server also sends live transcripts of the original speech and the translated text, for example to show captions.
eagerness controls the tradeoff between speed and quality. high starts translating soonest, and low waits for more of the sentence. The translated voice uses Eleven v4 Turbo by default, and you can switch to eleven_v3 or eleven_flash_v2_5 with tts_model_id. For sensitive conversations, set enable_zrm=true to use Zero Retention Mode.
Video to Music
Video to Music writes a soundtrack for your video. The Video to music API is now available: you upload a video and get back music that fits it, without writing a music prompt from scratch. Use it to score product videos, ads and social clips.
You can send up to 10 video files (200 MB combined). They’re joined in order and scored as one video. To steer the result, add a text description and up to 10 style tags.
Three options are worth knowing. use_audio_as_reference takes the genre, BPM, instruments and mood from the video’s own audio. add_sfx adds sound effects that match the video. separate_tracks returns the music and the sound effects as separate files in a ZIP, so you can mix them yourself.
ElevenAgents
- Alerts: Alerts notify a workspace when agent error rates exceed a threshold or rise sharply above their recent baseline. Monitors cover conversation failures, tool failures, webhook errors, guardrail triggers, call initiation errors and evaluation success scores, and run with default thresholds without any setup. Agent-level thresholds and notifiers can be set through
platform_settings.alertingon Update agent. Notifications can be delivered as webhooks, to Slack or to PagerDuty. - Guardrails generally available: Guardrails are no longer in alpha.
- Workspace environments: Environments for environment variables can now be managed through the API with List environments, Create environment and Delete environment. A variable without a value for an environment falls back to its
productionvalue. Deleting an environment removes its values from all variables, andproductioncannot be deleted. - Answering machine detection actions: Twilio answering machine detection settings on Outbound call and Submit batch call accept
actions, a list of rules that pair anaction(hang_up) with theanswered_byverdicts that trigger it, such asmachine_startorfax. Verdicts without a rule leave the call running. - New LLMs: Agents can use
gemini-4-argon,claude-opus-4-6and the ElevenLabs-hostednemotron-35-super-vl-nvfp4. See Models. - Authored-only workflow reads: Get agent and Update agent accept
include_procedure_generated_workflow_elements. Set it tofalseto receive only the workflow nodes and edges you authored, excluding the ones compiled from structured procedures. Defaults totrue, so existing callers are unchanged. - Branch rebasing from any branch: Rebase branch and Preview rebase accept an optional
source_branch_idto pull changes from a branch other than main. - Delete agents used as subagents: Delete agent accepts
force. Whentrue, the agent is deleted even if other agents use it as a subagent, and it is removed from those agents and their branches. - Sort agents by last update: List agents accepts
sort_by=updated_at, and each agent summary includesupdated_at_unix_secs, the last time the main branch or agent-level settings changed. - Google Calendar event updates: The Google Calendar integration adds
google_calendar_update_eventandgoogle_calendar_delete_eventtools. Both are privileged tools that can change or cancel any event on the connected calendar.
Music and Sound Effects
- Zero Retention Mode: Music generation, composition plan, stem separation and video to music endpoints, plus Sound Effects, accept
enable_logging=falseto use Zero Retention Mode. Zero Retention Mode is available to enterprise customers. Music requests that setstore_for_inpainting=trueare rejected in this mode. - WAV output for sound effects: Sound Effects accepts
wav_8000,wav_16000,wav_22050,wav_24000,wav_32000,wav_44100andwav_48000output formats.
Speech Engine
- 40-minute conversations: Speech Engine supports conversations of up to 40 minutes. The default limit remains 10 minutes. Raise
conversation.max_duration_secondsto allow longer conversations. - Sort by last update: List Speech Engines accepts
sort_by=updated_at, and each summary includesupdated_at_unix_secs.
Dubbing
- Subtitles for dubbed languages: Dubbing project language outputs (
GET /v1/dubbing/project/{project_id}/language/{language_id}) includesubtitles, a signed URL to the translated lines as WebVTT subtitles timed to the source speech. The URL expires one hour after it is issued.
Voices
- Resubmittable voice reviews: Voice sharing
review_statusaddsdeclined_resubmittablefor declined voices that can be resubmitted for review.
Workspaces
- Disable service account API keys: Disable service account API key disables a service account API key without changing its other settings. It remains available after a workspace downgrades from a plan that includes service accounts.
- Group concurrency limits: List workspace groups and List service accounts return
group_concurrency_limit,group_dubbing_concurrency_limitandgroup_music_concurrency_limitfor each group, and group permissions addall_resources_access.
SDK releases
JavaScript SDK
- v2.74.0 - Added
client.conversationalAi.realtime.createClientSecretfor the Realtime API, with theRealtimeModeland OpenAI-compatible session configuration types, and anenvironmentsclient withlist,createanddelete. Also addsforceon agent deletion,sourceBranchIdon branch rebase and rebase preview,includeProcedureGeneratedWorkflowElementson agent reads,enableLoggingon Music and Sound Effects methods, WAV output formats for Sound Effects, service account API keydisable, and Twilio answering machine detectionactions. - v2.73.0 - Added
createAndWaithelpers for Flows image, video and Text to Speech generations and template runs (client.flows.templates.runs.createAndWait). They create the request and poll until it completes or fails, pacing requests with the server’sRetry-Afterheader. Afailedresult is returned rather than thrown, andwaitTimeoutInSecondsthrowsFlowsWaitTimeoutErrorif the work is not finished in time. The regenerated client adds the Video to musicadd_sfx,separate_tracksanduse_audio_as_referenceoptions.
Python SDK
- v2.74.1 -
music.compose_detailednow accepts and forwardsenable_logging, which the custom wrapper previously dropped, so Zero Retention Mode works with this method. - v2.74.0 - Added
client.conversational_ai.realtime.create_client_secretfor the Realtime API and anenvironmentsclient, with the same regenerated API changes as JavaScript SDK v2.74.0. - v2.73.0 - Added
create_and_waithelpers for Flows image, video and Text to Speech generations and template runs, for both sync and async clients, withwait_timeout_in_secondsandmin_poll_interval_in_secondsoptions and aFlowsWaitTimeoutErrorexported fromelevenlabs.flows_custom. Type checkers such as pyright now resolveclient.webhooks.construct_event,client.speech_to_text.realtime,client.text_to_speech.convert_realtimeand the custom Music methods. Includes the same regenerated API changes as the JavaScript SDK. - v2.73.1 - Fixed the test suite on Python 3.13 and later, and added Python 3.13 and 3.14 to CI.
ElevenLabs CLI
- v1.7.0 - Regenerated API commands, adding environments, realtime client secrets and service account API key disabling.
- v1.6.0 - Added
create-and-waitcommands forflows image,flows video,flows text-to-speechandflows templates runs. Each command returns a finished result, with--wait-timeoutand--min-poll-intervaloptions. - v1.5.1, v1.5.2, v1.5.3 and v1.6.1 - Bug fixes for onboarding, including Windows support.
Packages
- @elevenlabs/client@1.28.0 - Added the experimental
persistentSessionoption for resumable text conversations over WebSocket, and theonConversationHistorycallback for the transcript the server replays on resume. WebSocket sessions now also emitonConversationMetadata. - @elevenlabs/react@1.18.0 - Forwards the experimental
onConversationHistorycallback to@elevenlabs/client. - @elevenlabs/types@0.25.0 - Added the
conversation_historyserver event and thepersistent_session_tokenfield onconversation_initiation_metadata. - @elevenlabs/react-native@1.2.30 - Updated to
@elevenlabs/client@1.28.0and@elevenlabs/react@1.18.0. - @elevenlabs/convai-widget-core@0.20.0 and @elevenlabs/convai-widget-embed@0.20.0 - Added the
persistent-sessionattribute. Text conversations survive a dropped connection or a page reload: the widget keeps the transcript, reopens the chat, renders the history replayed by the server and resumes the conversation. End chat discards the stored conversation. Workspaces without persistent sessions fall back to a standard conversation. - @elevenlabs/convai-widget-core@0.19.1 and @elevenlabs/convai-widget-embed@0.19.1 - First message quick replies are shown below the first message in voice calls.
API
View API changes
New endpoints
ElevenAgents
-
Create realtime client secret -
POST /v1/convai/realtime/client_secrets- Create an ephemeral client secret for browser-based Realtime API sessions. Accepts an optionalsession(OpenAI-compatible session configuration withmodelofeleven_realtime_v1_mini(default) oreleven_realtime_v1_max,instructions,audio,tools,turn_detectionandmax_duration_secondsfrom 60 to 7,200, default 600) and optionalexpires_afterwith requiredseconds(integer). Returnsvalue(string),expires_at(integer) and the echoedsession. -
List environments -
GET /v1/convai/environments- List the workspace environments, includingproduction. Each environment hasname, nullableupdated_at_unix_secsand nullablecreated_by_user_id. -
Create environment -
POST /v1/convai/environments- Create an environment with a required, uniquename(string). Returns409if the name is taken. -
Delete environment -
DELETE /v1/convai/environments/{environment}- Delete an environment and remove its values from all environment variables. Returns204.
Workspaces
- Disable service account API key -
POST /v1/service-accounts/{service_account_user_id}/api-keys/{api_key_id}/disable- Disable a service account API key without changing its other settings.
Updated endpoints
ElevenAgents
- Get agent, Update agent, Preview rebase and Preview merge
- Added optional
include_procedure_generated_workflow_elements(boolean, defaulttrue). Whenfalse, the returned workflow contains only authored nodes and edges, without those generated from structured procedures.
- Added optional
- Create agent, Get agent and Update agent
- The
llmenum addsgemini-4-argon,claude-opus-4-6andnemotron-35-super-vl-nvfp4. - TTS configuration
effectsremoves the deprecatedseedfield.
- The
- Delete agent - Added optional
force(boolean, defaultfalse). - List agents -
sort_byaddsupdated_at. Agent summaries add requiredupdated_at_unix_secs(integer). - Rebase branch and Preview rebase - Added optional
source_branch_id(string). Defaults to the agent’s main branch. - Outbound call, Submit batch call and Get batch call
- Twilio machine detection configuration adds
actions, an array of rules with requiredaction(hang_up) and requiredanswered_by(array ofhuman,fax,unknown,machine_start,machine_end_beep,machine_end_silenceormachine_end_other). Each action can appear in at most one rule.
- Twilio machine detection configuration adds
- Get conversation details
- Transcript tool calls add nullable
tool_id(string). It is not set for tools without an ID, such as system tools. - Turn metrics add nullable
voice_id(string). - Tool results add nullable
to_agentandto_node(string). - WhatsApp conversation info adds nullable
recipientwith nullablephoneandbsuid(string), and nullablecall_permission_requestwith requiredstate(blocked,sent,acceptedordeclined) and nullablerequested_at_unix(integer). - The platform charge category enum adds
v3_addon,v4_addonandrag_addon.
- Transcript tool calls add nullable
- Get ticket, List tickets and List workspace tickets
- Ticket responses add required
status_changes, an array of objects withstatus, nullablechanged_by_user_id(string) andchanged_at_unix_secs(integer).
- Ticket responses add required
Text to Speech and Text to Dialogue
- Create speech, Stream speech, Create dialogue and Stream dialogue, including their timestamp variants -
use_pvc_as_ivcis now nullable and no longer declares a default offalse.
Music
- Compose music, Compose detailed, Stream music, Compose detailed stream, Separate stems, Video to music and
POST /v1/music/plan- Added optional
enable_loggingquery parameter (boolean, defaulttrue). Set it tofalsefor Zero Retention Mode.
- Added optional
- Video to music - Added optional
add_sfx,separate_tracksanduse_audio_as_reference(boolean, defaultfalse).
Sound Effects
- Create sound effect
- Added optional
enable_loggingquery parameter (boolean, defaulttrue). output_formataddswav_8000,wav_16000,wav_22050,wav_24000,wav_32000,wav_44100andwav_48000.
- Added optional
Speech Engine
- List Speech Engines -
sort_byaddsupdated_at. Summaries add requiredupdated_at_unix_secs(integer).
Speech to Text
- Create transcript and Get transcript - Edited transcripts add required, read-only
edited_text(string), matchingtextand theedited_textfield of the realtimeedited_transcriptevent.
Dubbing
GET /v1/dubbing/project/{project_id}/languageandGET /v1/dubbing/project/{project_id}/language/{language_id}- Language outputs add nullablesubtitles(string, signed URL).
Voices
- Get voice, List voices and Get voices - Voice sharing
review_statusaddsdeclined_resubmittable.
Workspaces
- List workspace groups and List service accounts
- Groups add nullable
group_concurrency_limit,group_dubbing_concurrency_limitandgroup_music_concurrency_limit(integer orunlimited). - Group permissions add
all_resources_access.
- Groups add nullable
ElevenAgents
- Merge and search triage tickets: Merge tickets folds up to 50 open tickets from the same agent into one target ticket. Merged tickets keep the status
mergedand reference the target throughmerged_into_ticket_id. List tickets and List workspace tickets add a case-insensitivesearchparameter that matches ticket titles, descriptions, comments and turn comments, and each result includes asearch_matchexcerpt showing where the query matched. - Workflow End node outcomes: Workflow End nodes accept an
outcomeofsuccess(default) orfailure, so a workflow can record whether a conversation ended successfully. The conversation error type addsworkflow_end_failure. - Call queueing enabled for new agents: New agents are created with call queueing enabled unless
platform_settings.queueing_config.enabledis set explicitly. Existing agents keep their current setting. - Post-call audio webhook retries: When Enable retries is on,
post_call_audiowebhooks are retried on the same schedule as transcription webhooks. Audio payloads over 50 MiB are not retried, and queued audio retries are capped at 400 MiB per webhook. - Strict tool overrides: The conversation override
tool_ids_strictmakes an overriddentool_idslist the complete set of tools available in the conversation, including tools on workflow nodes. It is ignored when set on the agent itself. - Widget end confirmation: The widget can ask users to confirm before ending a chat or call. Enable it with
end_confirmation_enabledin the widget configuration or theend-confirmationattribute.
Text to Speech
- Audio effects: Text to Speech and Text to Dialogue requests accept an optional
audio_effectsobject. It applies a filter preset, perceived distance, a reverb environment, a looped background noise bed and stereo panning to the generated audio. List valid preset, environment and background noise IDs withGET /v1/audio-effects/catalog. Reverb and non-zero panning produce stereo audio and are rejected forulaw_*,alaw_*andpcm_*output formats. Effects do not change the cost of a request.
Speech to Text
- Turn prediction commit strategy: The realtime Speech to Text WebSocket adds
turn_predictionas acommit_strategyvalue. In addition to silence-based segmentation, it commits a segment as soon as the model predicts the speaker has finished their turn. It is the default for models that predict turns, and partial transcripts from those models includeturn_probability. See transcripts and commit strategies.
Voices
- Voice collections API: Voice collections can now be managed through the API. Create, list, update and delete collections, and add or remove voices. See List voice collections.
Workspaces
- Basic seat fallback: Update member accepts
allow_downgrade_to_basic_seat. When unlocking a user whose original seat type has no seats available, the user is moved to a basic seat instead of the request failing. - Workspace ID on user: Get user returns the
workspace_idof the workspace the API key belongs to.
SDK releases
JavaScript SDK
- v2.72.0 - Added clients for voice collections, the audio effects catalog and triage ticket merging. Text to Speech and Text to Dialogue methods accept
audio_effects, triage ticket list methods add search and filters, and the realtime Speech to Text client adds theturn_predictioncommit strategy.
Python SDK
- v2.72.0 - Added clients for voice collections, the audio effects catalog and triage ticket merging. Text to Speech and Text to Dialogue methods accept
audio_effects, triage ticket list methods add search and filters, and the realtime Speech to Text client adds theturn_predictioncommit strategy.play()andsave()now accept any iterable of audio chunks, such as a list or tuple.RealtimeEvents.CLOSEhandlers receive a dictionary with the WebSocket closecodeandreason; handlers without parameters continue to work.
ElevenLabs CLI
- v1.5.0 - Regenerated API commands, adding voice collections, the audio effects catalog, triage ticket merging and agent merge proposals.
Packages
- @elevenlabs/client@1.27.0 and @elevenlabs/react@1.17.0 - Added a
previousTextoption toScribe.connect()anduseScribe()in microphone mode. The text is sent with the first audio chunk to give the model context, such as existing document text, for casing, punctuation and sentence continuation.MediaDeviceOutputnow receiveslibsampleratePath, so output resampling uses the same custom libsamplerate bundle path asMediaDeviceInput. - @elevenlabs/react-native@1.2.29 - Updated to
@elevenlabs/client@1.27.0and@elevenlabs/react@1.17.0. - @elevenlabs/convai-widget-core@0.19.0 and @elevenlabs/convai-widget-embed@0.19.0 - Added an optional confirmation dialog before ending a chat or call, enabled with the
end-confirmationattribute orend_confirmation_enabledconfig, with customizable text. Added carousel rich content with scrollable item cards that can include an image, subtitle, description and actions. Carousels remain in the transcript after a reply, without their actions. Rich content is now shown during voice calls when the transcript is visible.
API
View API changes
New endpoints
ElevenAgents
- Merge tickets -
POST /v1/convai/triage-tickets/{agentqa_ticket_id}/merge- Merge open tickets into the target ticket. Requiressource_ticket_ids(string array, 1–50 items) and returns the target ticket.
Voices
- List voice collections -
GET /v1/voices/collections- List the voice collections the caller can access, sorted by title. - Create voice collection -
POST /v1/voices/collections- Create a collection with requiredtitleandicon(string, maximum 1,000 characters each). - Get voice collection -
GET /v1/voices/collections/{collection_id}- Retrieve a collection with itstitle,iconand the caller’spermission_on_resource(admin,editor,commenterorviewer). - Update voice collection -
PATCH /v1/voices/collections/{collection_id}- Update the optionaltitleoricon. - Delete voice collection -
DELETE /v1/voices/collections/{collection_id} - Add voice to collection -
POST /v1/voices/collections/{collection_id}/voice/{voice_id} - Remove voice from collection -
DELETE /v1/voices/collections/{collection_id}/voice/{voice_id}
Audio effects
GET /v1/audio-effects/catalog- List the availablefilter_presets,environmentsandbackground_noises. Each entry has an ID, anameand a nullabledescription.
Updated endpoints
ElevenAgents
- Create agent, Get agent and Update agent
- Workflow End nodes add
outcome(successorfailure, defaultsuccess). The field is required in responses. - Standalone workflow agent nodes and agent transfer configurations add nullable
branch_id(string) to target a specific branch of the destination agent. When unset, the transfer follows the target agent’s traffic deployment. Transfer tool results add aconfiguredbranch reason. - Prompt override configurations add
tool_ids_strict(boolean, defaultfalse). - TTS configuration
effectsaddpan(number, -1 to 1, default0). Theseedfield is deprecated and has no effect.
- Workflow End nodes add
- Get widget config - Added
end_confirmation_enabled(boolean, defaultfalse). - List tickets
- Added optional
search(string, maximum 200 characters) andmerged_into_ticket_id(string) query parameters. - Ticket responses add nullable
merged_into_ticket_idand nullablesearch_matchwithfield(title,description,commentorturn_comment),snippet,highlight_start,highlight_endand nullableturn_index.
- Added optional
- List workspace tickets - Added optional
agent_id,sources,priorities,sort_by(created_atorpriority, defaultcreated_at),owner_user_id,issue_type,label,merged_into_ticket_idandsearchquery parameters. - List tests - Added optional
created_by_user_id(string). Use@mefor the authenticated user. - List test invocations - Added optional
test_id(string),version_id(string),start_timeandend_time(integer, Unix seconds, inclusive) query parameters. - Update agent branch - Added optional
auto_rebase(boolean). Branch responses addauto_rebase(boolean, defaultfalse). - Get conversation details
- Twilio, SIP trunk and Exotel phone call metadata add nullable
call_resultwithanswered,voicemail,busy,no_answerandfailedvalues. - Batch call metadata adds nullable
cold_attempt(integer) and nullableconversation_type(initial,retry,callbackorredirect). - The conversation error type adds
workflow_end_failure.
- Twilio, SIP trunk and Exotel phone call metadata add nullable
- Knowledge base auto-sync
auto_discoveris deprecated and no longer has any effect.
Text to Speech
- Create speech, Stream speech, Create speech with timing and Stream speech with timing
- Added optional nullable
audio_effectswithfilter_preset_id(string),distance(number, 0–1, default0),environment_id(string),background_noise_id(string),send_level(number, 0–1, default1) andpan(number, -1 to 1, default0). - Unknown effect IDs return
404. Stereo effects return400forulaw_*,alaw_*andpcm_*output formats.
- Added optional nullable
Text to Dialogue
- Create dialogue, Stream dialogue, Create dialogue with timestamps and Stream dialogue with timestamps - Added optional nullable
audio_effects, matching the Text to Speech endpoints.
Speech to Text
- Realtime Speech to Text
- The
commit_strategyquery parameter addsturn_prediction, alongsidemanualandvad. partial_transcriptmessages add optionalturn_probability(number), sent only for models that predict turns.
- The
Music
POST /v1/music/video-to-music- Responses add asong-idheader with the ID of the generated song.
History
- List generated items - The history item
sourceenum addsVoicePreview.
Workspace and user
- Update member - Added optional
allow_downgrade_to_basic_seat(boolean, defaultfalse). - Get user - Added required
workspace_id(string). - Service account API key permissions add
voice_collections_readandvoice_collections_write.
ElevenAgents
- Structured procedures compile on save and publish: Update agent and Create agent draft now compile and validate structured procedures automatically. Integrations no longer need to call the compile endpoint or manually update the generated workflow. Invalid procedures return per-procedure errors as
400 procedure_validation_failed. Agents without structured procedures are unaffected. - Compile procedures endpoint is legacy: Compile procedures is marked legacy and should not be used by new integrations. It remains available for existing callers.
- Merge proposals: Merge proposals are available in alpha. Teams can open a proposal for an agent branch, review configuration changes and test results, leave comments or approval decisions, and merge approved changes into a target branch. See Merge proposals.
SDK releases
JavaScript SDK
- v2.71.0 - Added clients for agent merge proposals, deployment history, test invocation cancellation and Speech Engine duplication. The release also adds triage ticket priorities and sorting, batch call export limits,
gpt-6.1-sol,claude-sonnet-5-5, Eleven v4 agent voice models and current agent configuration types.
Python SDK
- v2.71.0 - Added clients for agent merge proposals, deployment history, test invocation cancellation and Speech Engine duplication. The release also adds triage ticket priorities and sorting, batch call export limits,
gpt-6.1-sol,claude-sonnet-5-5, Eleven v4 agent voice models and current agent configuration types.
Swift SDK
- v3.4.0 - Dynamic variables now preserve native JSON types when sent in conversation initiation data.
Packages
- @elevenlabs/client@1.26.0, @elevenlabs/react@1.16.0, @elevenlabs/react-native@1.2.28 and @elevenlabs/types@0.24.0 - Added realtime Speech to Text transcript editing through
transcriptEdit, theedited_transcriptevent and React’sonEditedTranscriptcallback. React now forwardsfilterBackgroundAudioto realtime sessions. The client also exportsMessagePayloadand uses stable response IDs to prevent duplicate agent replies. - @elevenlabs/convai-widget-core@0.18.3 and @elevenlabs/convai-widget-embed@0.18.3 - Added image previews and download links for agent message attachments, including attachment-only replies. Stable response IDs prevent duplicate replies, and text-only conversations now use a stop control with an End chat tooltip.
API
View API changes
New endpoints
ElevenAgents
- Create a merge proposal -
POST /v1/convai/agents/{agent_id}/merge-proposals- Create a proposal with requiredsource_branch_id,target_branch_idandtitlefields, plus an optional description, reviewer IDs and triage ticket ID. - List merge proposals -
GET /v1/convai/agents/{agent_id}/merge-proposals- List proposals with status, source branch, target branch and text filters, plus cursor pagination. - Get a merge proposal -
GET /v1/convai/agents/{agent_id}/merge-proposals/{merge_proposal_id}- Retrieve a proposal and its review, comment and outcome details. - Update a merge proposal -
PATCH /v1/convai/agents/{agent_id}/merge-proposals/{merge_proposal_id}- Update its title, description or requested reviewers, or close it without merging. - Review a merge proposal -
POST /v1/convai/agents/{agent_id}/merge-proposals/{merge_proposal_id}/reviews- Submit anapprovedorchanges_requesteddecision with an optional comment. - Comment on a merge proposal -
POST /v1/convai/agents/{agent_id}/merge-proposals/{merge_proposal_id}/comments- Add a Markdown comment of up to 4,096 characters. - Merge a merge proposal -
POST /v1/convai/agents/{agent_id}/merge-proposals/{merge_proposal_id}/merge- Merge the source branch and optionally control forced conflict resolution and source branch archival. - List agent deployments -
GET /v1/convai/agents/{agent_id}/deployments- List current and historical branch traffic splits with page-based pagination. - Cancel a test invocation -
POST /v1/convai/test-invocations/{test_invocation_id}/cancel- Cancel a running test invocation.
Speech Engine
- Duplicate a Speech Engine -
POST /v1/speech-engine/{speech_engine_id}/duplicate- Duplicate a Speech Engine and optionally assign the copy a newname.
Updated endpoints
ElevenAgents
- Create agent, Get agent and Update agent
- Added
gpt-6.1-solandclaude-sonnet-5-5to theLLMenum. - Added
eleven_v4andeleven_v4_turboto the conversational TTS model options. The default changed fromeleven_flash_v2toeleven_v4_turbo. - Agent transfers and standalone workflow agent nodes add optional
preserve_voice_settings(boolean, defaultfalse) to retain the base voice, TTS model and voice settings after a transfer. - Skip-turn tool configurations add optional
wait_timeout_secs(number, default-1). A value of-1keeps the agent silent until the caller speaks. - Post-call webhook configurations add optional
exclude_transcript(boolean, defaultfalse) to omit turn-by-turn transcripts from JSON transcription webhooks. - Conversation history redaction configurations add optional
excluded_data_collection_ids(string array, maximum 50) to preserve selected extracted values while redacting their rationales. - Client tool
response_timeout_secsnow accepts-1to wait indefinitely whenexpects_responseis enabled.
- Added
- List tickets
- Added optional
priorities(array) withlow,medium,highandurgentvalues. - Added
sort_bywithcreated_atandpriorityvalues; the default iscreated_at. - Ticket responses now require
title, nullablepriority, and apriority_changesarray. Ticket creation and update requests can setpriority, and ticket status addscancelled.
- Added optional
- List tools - Added optional nullable
used_by_agent_id(string) to return tools referenced by an agent’s main branch. - Export batch call results - Added optional nullable
limit(integer, minimum1) to export only the first recipients as a column preview. - List test invocations and Get test invocation
- Test invocation summaries add required
cancelled_count(integer) and optionalcancelled(boolean, defaultfalse). - Test run status adds
cancelled.
- Test invocation summaries add required
- Get conversation details - Transcript entries add optional nullable
platform_eventwithuser_turn_timeoutandskip_turn_wait_elapsedvalues. - Get agent version metadata - Added optional nullable
merged_authors, an array describing the people who published changes merged into the main branch. - Get crawl job - Removed the deprecated
max_depthresponse field.
Eleven v4 et Eleven v4 Turbo
Eleven v4 et Eleven v4 Turbo sont désormais disponibles. Eleven v4 offre une synthèse vocale de meilleure qualité et plus expressive, ainsi qu’un clonage de voix amélioré dans plus de 90 langues. Eleven v4 Turbo propose la même famille de modèles pour les applications en temps réel, avec une latence d’inférence médiane d’environ 100 ms.
Utilisez eleven_v4 via l’API Text to Dialogue pour la création de contenu et les contenus audio longs. Utilisez eleven_v4_turbo via le WebSocket Text to Dialogue pour les agents et les applications interactives.
Speech to Text
- Modification des transcriptions : la transcription par lot et en temps réel accepte désormais une instruction de modification en langage naturel pouvant contenir jusqu’à 2 000 caractères. Les réponses par lot renvoient le résultat dans
edited_transcript, tandis que les sessions en temps réel émettent un événementedited_transcript. Consultez les guides par lot et en temps réel pour connaître les incompatibilités, la gestion des réponses et la tarification.
ElevenAgents
- Modèles d’agent : ajout de
gpt-6-sol,gpt-6-luna,glm-52,deepseek-v41-flash,claude-opus-5etclaude-opus-5-5aux options de LLM d’agent prises en charge. - Utilisation des tests d’agent : les résumés d’invocation de test peuvent inclure le total de crédits et le prix en USD. Les exécutions de test individuelles peuvent inclure les crédits et le détail de la facturation.
Flows
- API de modèles : ajout de points de terminaison pour lister et récupérer les modèles Flows publiés, démarrer des exécutions de modèles et récupérer les résultats d’exécution. Les exécutions peuvent cibler une version publiée et envoyer leur résultat final à un webhook configuré.
Text to Dialogue
- Continuité de génération : les requêtes Text to Dialogue peuvent utiliser du texte précédent ou suivant et des identifiants de requête afin de maintenir la continuité entre les générations.
- Mode voix professionnelle : les requêtes Text to Dialogue peuvent utiliser la version Instant Voice Clone d’un Professional Voice Clone afin de réduire la latence ou d’améliorer l’expressivité.
Versions des SDK
SDK JavaScript
- v2.70.0 - Ajout des types de modification de transcription Speech to Text par lot et en temps réel, du filtrage
branchIdet des champs de coût pour les résumés d’invocation de test, de la découverte automatique de la base de connaissances, deusePvcAsIvcpour Text to Dialogue et du modèle d’agentdeepseek-v41-flash.
SDK Python
- v2.70.0 - Ajout des types de modification de transcription Speech to Text par lot et en temps réel, du filtrage
branch_idet des champs de coût pour les résumés d’invocation de test, de la découverte automatique de la base de connaissances, deuse_pvc_as_ivcpour Text to Dialogue et du modèle d’agentdeepseek-v41-flash.
CLI ElevenLabs
- v1.4.0 - Correction des requêtes utilisant une clé API explicite afin qu’elles n’envoient plus également les identifiants OAuth enregistrés, ce qui provoquait des échecs d’authentification. Ajout de descriptions pour le groupe de commandes de commentaires et d’instructions permettant de signaler les capacités manquantes aux compétences d’agent générées. Régénération du client API avec les points de terminaison de modèles Flows et les types de schéma actuels.
API
Voir les modifications de l’API
Nouveaux points de terminaison
Flows
- Lister les modèles -
GET /v1/flows/templates- Liste les modèles publiés et leurs versions exécutables. - Obtenir un modèle -
GET /v1/flows/templates/{template_id}- Récupère un modèle et ses versions exécutables. - Créer une exécution de modèle -
POST /v1/flows/templates/{template_id}/runs- Démarre une exécution de modèle avec des valeurs d’entrée, une version facultative et un webhook facultatif. - Lister les exécutions de modèles -
GET /v1/flows/templates/{template_id}/runs- Liste les exécutions de modèles avec pagination par curseur et filtrage facultatif par version. - Obtenir une exécution de modèle -
GET /v1/flows/templates/{template_id}/runs/{run_id}- Récupère une exécution de modèle et l’état de sa sortie.
Points de terminaison mis à jour
Speech to Text
- Convertir la parole en texte -
POST /v1/speech-to-text- Ajout de
transcript_editfacultatif et nullable (chaîne), qui accepte une instruction de modification en langage naturel pouvant contenir jusqu’à 2 000 caractères. - Les réponses ajoutent
edited_transcriptfacultatif et nullable. Une modification réussie contient les champs obligatoireskind: "transcript"ettext. Une modification échouée contient les champs obligatoireskind: "error",error_type: "edit_failed"etmessage.
- Ajout de
- Realtime Speech to Text accepte
transcript_edit(chaîne) et émet des événementsedited_transcriptavec les champs obligatoirestextetedited_text.
ElevenAgents
- Créer un agent, Obtenir un agent et Mettre à jour un agent
- Ajout de
gpt-6-sol,gpt-6-luna,glm-52,deepseek-v41-flash,claude-opus-5etclaude-opus-5-5à l’énumérationLLM. - Les moniteurs d’alerte ajoutent
enabledfacultatif et nullable (booléen) afin de contrôler si un moniteur peut envoyer des notifications.
- Ajout de
- Lister les conversations et Recherche textuelle
data_collection_paramsetdynamic_variable_paramsajoutent les opérateursneqetin. Les valeurs deinutilisent un séparateur de type barre verticale.evaluation_paramsacceptesuccess,failureouunknowncomme valeur de résultat.
- Recherche intelligente -
GET /v1/convai/conversations/messages/smart-search- Ajout de
branch_idfacultatif et nullable (chaîne).
- Ajout de
- Lister les invocations de test -
GET /v1/convai/test-invocations- Ajout de
branch_idfacultatif et nullable (chaîne). - Les résumés d’invocation de test ajoutent
credits_usedfacultatif et nullable (entier) ettotal_price(nombre).
- Ajout de
- Obtenir une invocation de test -
GET /v1/convai/test-invocations/{test_invocation_id}- Les exécutions de test individuelles ajoutent
credits_usedfacultatif et nullable (entier) etcharging(ConversationChargingCommonModel).
- Les exécutions de test individuelles ajoutent
- Simuler une conversation et Diffuser une simulation de conversation
- Modification de la valeur par défaut facultative de
new_turns_limit, de10,000à100, et définition de sa valeur maximale à100.
- Modification de la valeur par défaut facultative de
- Créer une tâche d’exploration et Obtenir une tâche d’exploration
- Ajout de
auto_discoverfacultatif (booléen, valeur par défautfalse) pour suivre les liens et ajouter les nouvelles pages découvertes lors de la synchronisation automatique. Nécessiteenable_auto_sync: true.
- Ajout de
- Actualiser un document de la base de connaissances -
POST /v1/convai/knowledge-base/{documentation_id}/refreshAutoSyncInfoajouteauto_discoverfacultatif (booléen, valeur par défautfalse) afin d’indiquer si les actualisations explorent et ajoutent les nouvelles pages découvertes.
- Lister les outils du serveur MCP -
GET /v1/convai/mcp-servers/{mcp_server_id}/tools- Les réponses ajoutent
tool_approval_statusesfacultatif (tableau). Chaque élément contienttool_idobligatoire (chaîne) etstate(up_to_date,needs_reviewounot_approved), avec des champs facultatifs de politique d’approbation et de définition approuvée.
- Les réponses ajoutent
Text to Dialogue
- Convertir du texte en dialogue, Diffuser du texte en dialogue, Convertir avec horodatages et Diffuser avec horodatages
- Ajout de
use_pvc_as_ivcfacultatif (booléen, valeur par défautfalse) pour utiliser la version Instant Voice Clone d’un Professional Voice Clone. - Ajout de
previous_textetfuture_textfacultatifs et nullables (chaînes, maximum 100 caractères), ainsi que deprevious_request_idsetnext_request_ids(tableaux de chaînes, maximum 3 identifiants) pour assurer la continuité de génération. - L’objet
settingsfacultatif ajoutesimilaritynullable (nombre, 0–1, valeur par défaut0.75).
- Ajout de
Text to Speech
- Convertir du texte en parole, Diffuser du texte en parole, Convertir avec horodatages et Diffuser avec horodatages
- Le champ
use_pvc_as_ivcexistant n’est plus obsolète.
- Le champ
Flows
- Créer une génération d’image, Créer une génération vidéo et Créer une exécution de modèle
- Les cibles de webhook nécessitent désormais un discriminateur
typeexplicite (allouids). Les valeurs par défaut précédentes ont été supprimées.
- Les cibles de webhook nécessitent désormais un discriminateur
Voice Design
- Créer une voix à partir d’un aperçu, Créer des aperçus de voix, Concevoir une voix et Remixer une voix
- Ajout d’une réponse
409lorsqu’une génération pour l’aperçu est déjà en cours. Réessayez la requête après avoir reçu cette réponse.
- Ajout d’une réponse
Workspaces
- Créer une clé API de compte de service -
POST /v1/service-accounts/{service_account_user_id}/api-keys- Ajout de
tts_concurrency_limit,music_concurrency_limitetdubbing_concurrency_limitfacultatifs et nullables (entiers) pour les clés API de compte de service Enterprise.
- Ajout de
- Mettre à jour une clé API de compte de service -
PATCH /v1/service-accounts/{service_account_user_id}/api-keys/{api_key_id}- Ajout de
tts_concurrency_limit,music_concurrency_limitetdubbing_concurrency_limitfacultatifs. Chaque champ accepte un entier,clear,no_updateounull, et utiliseno_updatepar défaut.
- Ajout de
Modèles
- Lister les modèles -
GET /v1/models- Dépréciation de
max_characters_request_free_useretmax_characters_request_subscribed_user, auparavant obligatoires. Ces champs ne sont pas appliqués, utilisez plutôtmaximum_text_length_per_request.
- Dépréciation de
Image & Vidéo
- Sora 2 et Sora 2 Pro retirés : OpenAI met fin à l’API Sora le 24 septembre 2026. Les deux modèles ont été retirés du sélecteur de modèles Image & Vidéo et cesseront de générer à cette date. Les générations existantes restent disponibles dans votre historique. Les Flows et modèles qui utilisent un nœud Sora doivent être remplacés par un autre modèle vidéo, tel que Gemini Omni 1.1 Flash, avant de pouvoir être exécutés à nouveau.
- Seedance 1.5 Pro déprécié : ByteDance retire Seedance 1.5 Pro le 11 novembre 2026. Le modèle n’est plus proposé pour les nouvelles générations et cessera de fonctionner à cette date. Utilisez plutôt un modèle Seedance 2.0.
ElevenAgents
- Appels d’outils parallèles : la configuration des prompts de l’agent inclut désormais
enable_parallel_tool_calls(booléen, valeur par défauttrue). Lorsqu’il est activé, les modèles compatibles peuvent exécuter plusieurs outils au cours d’un même tour. - Modèles d’agent : ajout de
gpt-6-astraaux options LLM des agents. - Alertes : les alertes d’agent prennent désormais en charge les canaux Slack en plus des notifications PagerDuty et webhook.
- Recherche de numéros de téléphone : ajout d’un point de terminaison de numéros de téléphone paginé par curseur, avec des filtres pour le fournisseur, la prise en charge des appels sortants, l’agent et la branche attribués, le libellé et le numéro de téléphone.
Image et vidéo
- Modèles GPT Image 2.5 : la génération d’images prend désormais en charge
gpt-image-2.5-flareetgpt-image-2.5-sunburst. Les deux modèles acceptent jusqu’à 10 images de référence, des niveaux de qualité jusqu’àmax, 14 formats d’image fixes ainsi queauto, et une sortie en 1K, 2K ou 4K.
Speech Engine
- Délai d’expiration en cascade : les requêtes de création et de mise à jour de Speech Engine acceptent désormais
cascade_timeout_seconds(nombre, 2–15, valeur par défaut4) afin de contrôler le temps pendant lequel ElevenLabs attend un moteur de synthèse vocale en amont avant de réessayer.
Text to Speech
- Limites de tours multi-contexte : le marqueur
is_final_audio_for_turnest désormais émis après chaque octet mis en mémoire tampon pour ce tour, y compris pour les sorties MP3 et Opus. Les clients peuvent utiliser ce marqueur comme limite de tour exacte sans passer au PCM.
Versions des SDK
SDK JavaScript
- v2.69.0 - Ajout de
transcriptEditau wrapper Speech to Text en temps réel, avec du texte modifié transmis via les événementsedited_transcript. Correction de la vérification JWT de Speech Engine pour les clés API avec suffixes de résidence des données. Ajout dephoneNumbers.listV2et de méthodes de modèles Flows pour lister et obtenir des modèles, ainsi que lister, créer et obtenir des exécutions. Clients et types régénérés pour GPT Image 2.5, les alertes d’agent Slack, les appels d’outils parallèles,gpt-6-astra, les limites de requêtes simultanées des clés API de comptes de service, le mode zéro rétention de Music, les filtres de conversation et les statuts d’approbation des outils MCP.
SDK Python
- v2.69.0 - Ajout de
transcript_editau wrapper Speech to Text en temps réel, avec du texte modifié transmis via les événementsedited_transcript. Correction de la vérification JWT de Speech Engine pour les clés API avec suffixes de résidence des données, et modification deplayafin de générer une erreur en cas d’échec deffplay. Ajout dephone_numbers.list_v2et de méthodes de modèles Flows pour lister et obtenir des modèles, ainsi que lister, créer et obtenir des exécutions. Clients et types régénérés pour GPT Image 2.5, les alertes d’agent Slack, les appels d’outils parallèles,gpt-6-astra, les limites de requêtes simultanées des clés API de comptes de service, le mode zéro rétention de Music, les filtres de conversation et les statuts d’approbation des outils MCP.
CLI ElevenLabs
- v1.3.2 - Ajout de métadonnées facultatives
--intentpour recueillir des retours sur l’utilisation de la CLI. Ajout deelevenlabs feedback missing-capabilitypour signaler les workflows que la CLI ne prend pas en charge.
API
Voir les modifications de l'API
Nouveaux points de terminaison
ElevenAgents
- Lister une page de numéros de téléphone -
GET /v1/convai/v2/phone-numbers- Ajout des paramètres de pagination facultatifs
page_size(entier, 1–1 000, valeur par défaut100) etcursor(chaîne pouvant être nulle). - Ajout des filtres facultatifs
search,label,phone_number,provider,supports_outbound,agent_id,branch_id,sort_byetsort_direction.sort_byacceptelabelouphone_number;sort_directionutiliseascpar défaut. - La réponse contient
phone_numbers(tableau) obligatoire,next_cursor(chaîne) facultatif et pouvant être nul, ainsi quehas_more(booléen, valeur par défautfalse).
- Ajout des paramètres de pagination facultatifs
Points de terminaison mis à jour
ElevenAgents
- Créer un agent, Obtenir un agent et Mettre à jour un agent
- La configuration des prompts ajoute
enable_parallel_tool_calls(booléen, valeur par défauttrue). Les remplacements de workflow acceptent un booléen pouvant être nul. - L’énumération
LLMajoutegpt-6-astra, y compris pour les remplacements d’initialisation de conversation. - Les notifications d’intégration d’alertes utilisent désormais
integration_typepour sélectionnerpagerdutyouslack. Les notifications Slack requièrentconnection_idetchannel_id(chaînes) ;channel_idaccepte 1 à 64 caractères.
- La configuration des prompts ajoute
- Lister les conversations et Résoudre une référence de conversation
data_collection_paramsn’accepte plus l’opérateurmissing. Utilisez une valeur vide pour trouver les conversations dans lesquelles le champ n’a pas été collecté.- Une valeur vide dans
dynamic_variable_paramscorrespond aux conversations dans lesquelles la variable n’a pas été définie.
- Créer un Speech Engine, Obtenir un Speech Engine et Mettre à jour un Speech Engine
- Ajout de
cascade_timeout_seconds(nombre, 2–15). Les requêtes de création utilisent4par défaut, les requêtes de mise à jour acceptent une valeur pouvant être nulle et les réponses requièrent ce champ.
- Ajout de
Image et vidéo
- Créer une génération d’image -
POST /v1/flows/image- Ajout de variantes de requête pour
gpt-image-2.5-flareetgpt-image-2.5-sunburst. - Les deux variantes requièrent
promptetmodel_id, acceptent jusqu’à 10imageset ajoutentresolution(1K,2Kou4K, valeur par défaut1K). qualityacceptelow,medium,high,xhighoumaxet utilisehighpar défaut.aspect_ratioaccepteautoet 14 ratios fixes, et utilise16:9par défaut.
- Ajout de variantes de requête pour
Doublage
- Créer un projet de doublage -
POST /v1/dubbing/projectkeytermsmultipartite accepte des champs répétés ou un seul tableau JSON.webhook_idsmultipartite accepte des champs répétés, un seul tableau JSON ou une chaîne séparée par des virgules.
- Obtenir une ressource de Workspace, Partager une ressource de Workspace et Ne plus partager une ressource de Workspace
WorkspaceResourceTypeajoutedubbing_project.
Base de connaissances
- Créer une tâche d’exploration -
POST /v1/convai/knowledge-base/crawl- Dépréciation de
max_depth(entier, valeur par défaut3). Ce champ n’a désormais aucun effet et n’est plus soumis à des contraintes minimales ou maximales. - Les métadonnées des tâches d’exploration ne requièrent plus
max_depth.
- Dépréciation de
Utilisateur
- Obtenir un utilisateur et Obtenir un abonnement
- Les énumérations de devises ajoutent
gbp.
- Les énumérations de devises ajoutent
Modifications du schéma
Text to Speech
- Les messages WebSocket
audioetfinalmulti-contexte définissent désormaisis_final_audio_for_turncomme intervenant après la livraison de tous les octets mis en mémoire tampon pour le tour.
ElevenAgents
- Mise en file d’attente des appels et audio d’attente : ajout de la mise en file d’attente des appels pour les agents ayant atteint leur limite de requêtes simultanées. Les appelants en attente entendent un audio d’attente et reçoivent des événements
queue_statusjusqu’à leur admission ou l’expiration du délai. De nouvelles méthodes API permettent d’importer ou de supprimer un audio d’attente personnalisé. - Métadonnées des versions de test : les résultats des tests d’agent identifient désormais la version de l’agent sur laquelle ils ont été exécutés et indiquent s’ils incluaient des modifications de brouillon. Les récapitulatifs d’invocation signalent également lorsque des soumissions partielles ont été exécutées sur différentes versions.
- Métadonnées de modèle et de compte : la configuration LLM de l’agent accepte désormais
gemini-3.8-flash. Les réponses des comptes WhatsApp indiquent si le compte utilise l’API Cloud ou le processus d’inscription en coexistence.
Musique
- Prise en charge de l’API Music v2.5 : les points de terminaison Music et les SDK acceptent désormais
music_v2_5. Les segments de génération de plans de composition prennent en charge jusqu’à 6 132 caractères, avec jusqu’à 30 lignes de 200 caractères chacune.
Versions des SDK
SDK JavaScript
- v2.68.0 - Ajout de méthodes pour importer et supprimer l’audio d’attente d’un agent. Types régénérés pour la mise en file d’attente des appels, les références de procédures, la création de branches de brouillon, les limites de requêtes RAG, les métadonnées de versions de test,
gemini-3.8-flashetmusic_v2_5.
SDK Python
- v2.68.0 - Ajout de méthodes synchrones et asynchrones pour importer et supprimer l’audio d’attente d’un agent. Types régénérés pour la mise en file d’attente des appels, les procédures, la création de branches de brouillon, les limites de requêtes RAG, les métadonnées de versions de test,
gemini-3.8-flashetmusic_v2_5. Le wrappercompose_detailedaccepte désormais chaque ID de modèle de musique généré ainsi que les plans de composition basés sur des prompts et sur des segments.
SDK iOS
- v3.3.1 - Rapproche les messages de l’agent par
response_idplutôt que parevent_id, en préservant plusieurs réponses autour des appels d’outils et en évitant que des parties diffusées en streaming ou des corrections soient associées au mauvais message.
CLI ElevenLabs
- v1.3.0 - Ajout de métadonnées facultatives
--intentaux commandes API générées. Ajout deelevenlabs feedback missing-capabilitypour signaler les workflows que la CLI ne prend pas en charge.
Packages
- @elevenlabs/convai-widget-core@0.18.2 et @elevenlabs/convai-widget-embed@0.18.2 - Correction des réponses d’agent en double après les appels d’outils et de procédures. Les réponses textuelles diffusées en streaming conservent leur mise en forme Markdown, et les conversations vocales n’affichent plus les parties de chat diffusées mais non prononcées. Le package embed limite également les événements
user_activityà un par seconde et supprime cette limite à la fin d’une session.
API
Voir les modifications de l'API
Nouveaux points de terminaison
ElevenAgents
- Importer l’audio d’attente d’un agent -
POST /v1/convai/agents/{agent_id}/hold-audio- Accepte le champ obligatoire
hold_audio_file(binaire) sous forme de données de formulaire multipartites. Le fichier doit être au format MP3 ou WAV, ne pas dépasser 40 Mo ni 180 secondes. - Renvoie les champs obligatoires
agent_id(chaîne) ethold_audio(AgentHoldAudioConfig). L’objet d’audio d’attente requiertaudio_path,audio_url,original_filename(chaînes),duration_secs(nombre) etsize_bytes(entier).
- Accepte le champ obligatoire
- Supprimer l’audio d’attente d’un agent -
DELETE /v1/convai/agents/{agent_id}/hold-audio- Supprime le clip personnalisé afin que les appelants en attente entendent la tonalité d’attente par défaut.
- Renvoie le champ obligatoire
agent_id(chaîne).
Points de terminaison mis à jour
ElevenAgents
- Créer un agent, Obtenir un agent et Mettre à jour un agent
- Ajout de
platform_settings.queueing_config(AgentQueueingConfig) facultatif, avecenabled(booléen, valeur par défautfalse) etwait_timeout_seconds(entier, 1–1 800, valeur par défaut180). - Les réponses de configuration de file d’attente incluent
hold_audio(AgentHoldAudioConfig) en lecture seule et pouvant être nul. Les réponses d’agent incluent égalementdefault_hold_audio_url(chaîne) facultatif.
- Ajout de
- Obtenir une conversation -
GET /v1/convai/conversations/{conversation_id}- Les métadonnées ajoutent
queue_wait_secs(nombre) facultatif et pouvant être nul. Le temps d’attente dans la file est exclu decall_duration_secset du temps facturé.
- Les métadonnées ajoutent
- Créer une branche d’agent -
POST /v1/convai/agents/{agent_id}/branches- Ajout de
include_draft(booléen, valeur par défautfalse) facultatif. Lorsqu’il est défini surtrue, une branche créée à partir de l’extrémité de la branche inclut l’ensemble des procédures de brouillon de l’appelant.
- Ajout de
- Lister les procédures et les points de terminaison de création, d’obtention et de brouillon des procédures
- Les éléments de liste ajoutent
referenced_tool_ids,referenced_kb_ids,referenced_procedure_idsetreferenced_dynamic_variables(tableaux de chaînes) facultatifs. - Les réponses d’agent et d’aperçu de fusion ajoutent
proceduresfacultatif, une map indexée par ID de procédure.
- Les éléments de liste ajoutent
- Lister les invocations de test -
GET /v1/convai/test-invocations- Ajout de
search(chaîne) facultatif et pouvant être nul pour filtrer les tests et les dossiers par nom.
- Ajout de
- Exécuter des tests sur un agent et Obtenir une invocation de test
- Les réponses d’invocation de suite de tests et d’exécution de test unitaire ajoutent
version_id(chaîne) facultatif et pouvant être nul, ainsi queran_against_draft(booléen, valeur par défautfalse) facultatif. - Les récapitulatifs d’invocation ajoutent également
runs_diverged_from_version(booléen, valeur par défautfalse) facultatif.
- Les réponses d’invocation de suite de tests et d’exécution de test unitaire ajoutent
POST /v1/convai/agents/{agent_id}/knowledge-base/rag-query- Le corps de la requête ajoute les remplacements spécifiques à la requête
max_documents_length(entier, 1–50 000) etmax_retrieved_rag_chunks_count(entier, 1–20), facultatifs et pouvant être nuls.
- Le corps de la requête ajoute les remplacements spécifiques à la requête
- Obtenir un compte WhatsApp -
GET /v1/convai/whatsapp-accounts/{phone_number_id}- Ajout de
account_type(WhatsAppAccountType, valeur par défautcloud_api) facultatif. Les valeurs sontcloud_apietcoexistence.
- Ajout de
- Calculer l’utilisation LLM attendue -
POST /v1/convai/agent/{agent_id}/llm-usage/calculateLLMUsageCalculatorLLMResponseModelajouteprice_per_message(nombre) obligatoire.
Musique
MusicModelIDajoutemusic_v2_5pourPOST /v1/music/video-to-music,POST /v1/music/plan, Composer de la musique,POST /v1/music/detailed,POST /v1/music/detailed/stream,POST /v1/music/stream,POST /v1/music/uploadetPOST /v1/music/finetunes.GenerationChunk.textaugmentemaxLengthde 6 000 à 6 132 caractères pourPOST /v1/music/plan,POST /v1/music,POST /v1/music/detailed,POST /v1/music/detailed/stream,POST /v1/music/streametPOST /v1/music/upload. Les noms de sections peuvent contenir 1 à 100 caractères, et les segments peuvent contenir jusqu’à 30 lignes de 200 caractères chacune.
Modifications du schéma
ElevenAgents
- Les événements serveur en temps réel ajoutent
queue_status. Son champ obligatoirequeue_status_event.statusacceptewaiting,admittedoutimed_out;timed_outprécède le code de fermeture WebSocket4300. L’audio d’attente continue d’arriver sous forme d’événementsaudioclassiques. - L’énumération
LLMajoutegemini-3.8-flash, y compris pour les remplacements d’initialisation de conversation. - Les unions de résultats des outils système de conversation ajoutent
start_procedure_success,start_procedure_error,end_procedure_successetend_procedure_error. FeaturesUsageCommonModelajoute les statuts de fonctionnalité facultatifsfreeform_procedureetstructured_procedure.