Models
Learn how to choose the right model for your use-case
The ElevenLabs Agents Platform provides a unified interface to connect your agent to multiple models and providers, offering flexibility, reliability, and cost optimization.
Key features
- Unified access: Switch between providers and models with minimal code changes
- High reliability: Automatically cascade from one provider to another if one fails
- Spend monitoring: Monitor your spending across different models
Supported models
Currently, the following models are natively supported and can be configured via the agent settings:
Pricing is typically denoted in USD per 1 million tokens unless specified otherwise. A token is a fundamental unit of text data for LLMs, roughly equivalent to 4 characters on average.
Custom LLM
Using your own custom LLM is supported by specifying the endpoint we should make requests to and providing credentials through our secure secret storage. Learn more about custom LLM integration.
With EU data residency enabled, a small number of older Gemini and Claude LLMs are not available in ElevenLabs Agents to maintain compliance with EU data residency. Custom LLMs and OpenAI LLMs remain fully available. For more information please see GDPR and data residency.
Choosing a model
Selecting the most suitable LLM for your application involves considering several factors:
- Task complexity: More demanding or nuanced tasks generally benefit from more powerful models (e.g., OpenAI’s GPT-4 series, Anthropic’s Claude Sonnet 4, Google’s Gemini 2.5 models)
- Latency requirements: For applications requiring real-time or near real-time responses, such as live voice conversations, models optimized for speed are preferable (e.g., Google’s Gemini Flash series, Anthropic’s Claude Haiku, OpenAI’s GPT-4o-mini)
- Context window size: If your application needs to process, understand, or recall information from long conversations or extensive documents, select models with larger context windows
- Cost-effectiveness: Balance the desired performance and features against your budget. LLM prices can vary significantly, so analyze the pricing structure (input, output, and cache tokens) in relation to your expected usage patterns
- HIPAA compliance: If your application involves Protected Health Information (PHI), it is crucial to use an LLM that is designated as HIPAA compliant and ensure your entire data handling process meets regulatory standards
The maximum system prompt size is 2MB, which includes your agent’s instructions, knowledge base content, and other system-level context.
Model configuration
Temperature
Temperature controls the randomness of model responses. Lower values produce more consistent, focused outputs while higher values increase creativity and variation.
- Low (0.0-0.3): Deterministic, consistent responses for structured interactions
- Medium (0.4-0.7): Balanced creativity and consistency
- High (0.8-1.0): Creative, varied responses for dynamic conversations
Backup LLM configuration
Configure backup LLMs to ensure conversation continuity when the primary LLM fails or becomes unavailable.
Configuration options:
- Default: Uses ElevenLabs’ recommended fallback sequence
- Custom: Define your own cascading sequence of backup models
- Disabled: No fallback (strongly discouraged for production)
Disabling backup LLMs means conversations will end abruptly if your primary LLM fails or becomes unavailable. This is strongly discouraged for production use.
Learn more about LLM cascading.
Thinking budget
Control how many internal reasoning tokens the model can use before responding. More tokens improve answer quality but slow down response time.
Options:
- Disabled: Fastest replies with no internal reasoning overhead
- Low: Minimal reasoning for quick responses
- Medium: Balanced reasoning and speed
- High: Maximum reasoning for complex queries
Reasoning effort
Some models support configurable reasoning effort levels (None, Low, Medium, High).
For conversational use-cases:
Keep reasoning effort set to None to avoid the agent thinking too long, which can disrupt natural conversation flow.
For workflow steps:
Reasoning effort is perfect for workflow steps that require complex thought or decision-making where response time is less critical.
Understanding pricing
- Tokens: LLM usage is typically billed based on the number of tokens processed. As a general guideline for English text, 100 tokens is approximately equivalent to 75 words
- Input vs. output pricing: Providers often differentiate pricing for input tokens (the data you send to the model) and output tokens (the data the model generates in response)
- Cache pricing:
input_cache_read: This refers to the cost associated with retrieving previously processed input data from a cache. Utilizing cached data can lead to cost savings if identical inputs are processed multiple timesinput_cache_write: This is the cost associated with storing input data into a cache. Some LLM providers may charge for this operation
- The prices listed in this document are per 1 million tokens and are based on the information available at the time of writing. These prices are subject to change by the LLM providers
For the most accurate and current information on model capabilities, pricing, and terms of service, always consult the official documentation from the respective LLM providers (OpenAI, Google, Anthropic).
HIPAA compliance
Certain LLMs available on our platform may be suitable for use in environments requiring HIPAA compliance, please see the HIPAA compliance docs for more details.
