Platform Models

The platform integrates with multiple language model providers and model families, giving you the flexibility to choose the right AI engine for your specific needs. Different models excel at different tasks - some are optimized for speed and efficiency, while others prioritize reasoning capability, creativity, or specialized knowledge domains.

Discovering Available Models

To see which language models are currently available on the platform and accessible with your account:

The response provides detailed information about each model including:

• id: Unique identifier for the model (e.g., "gpt-4", "claude-3-opus")
• description: Overview of the model's capabilities and characteristics
• provider: The organization providing the model (OpenAI, Anthropic, etc.)
• family: The model family or series it belongs to
• maxTokens: Total token capacity (input + output combined)
• maxInputTokens: Maximum tokens that can be sent to the model
• maxOutputTokens: Maximum tokens the model can generate in response

Understanding Token Limits

Token limits are critical when working with language models. Tokens are pieces of text that models process - roughly equivalent to words, though the exact tokenization varies by model. When building conversational applications, you need to account for:

1. Input tokens: Your prompt, system instructions, conversation history, and context
2. Output tokens: The model's generated response
3. Total tokens: The sum of input and output, which cannot exceed maxTokens

If you exceed token limits, the model will either truncate input or fail to process the request. Design your applications to manage context efficiently, potentially summarizing or truncating conversation history for long interactions.

Choosing the Right Model

Different models have different strengths:

• Large context models (high maxInputTokens): Ideal for processing lengthy documents or maintaining extensive conversation history
• Fast models (optimized for speed): Best for real-time chat applications requiring quick responses
• Reasoning models (optimized for accuracy): Suitable for complex analytical tasks, problem-solving, or technical questions
• Specialized models: Some models excel at code generation, creative writing, or specific language tasks

Model Availability and Visibility

The list endpoint only returns models that are visible and accessible based on your subscription plan. Some advanced or specialized models may require specific plan levels or additional agreements. Model availability may also change as new models are released or deprecated by providers.

Note: Model pricing, which varies by provider and model, will be added to the API response in a future update. For current pricing information, refer to the platform documentation or your account dashboard.