Models for agents

Agent model availability can differ from the direct inference catalog. Agent keys expose the models currently supported by the Agent API. Retrieve the list from GET https://llmbase.ai/api/v1/agents/models before selecting a model for OpenClaw, Hermes, or another compatible agent.

The response uses the OpenAI models.list() shape and includes LLMBase metadata such as pricing, supported_features, and prompt_cache_supported when available. If a model is not returned by this endpoint, it cannot be used with a chat agent key. Chat completions for unavailable agent models return 403 with model_not_available_for_agents.

The agent model view includes only chat/tool-capable models from LLMBase’s cost-controlled inference catalog. Image, embedding, rerank, classification, and non-agent models are not returned.

Using public model IDs

Agents always send a stable LLMBase model ID returned by GET https://llmbase.ai/api/v1/agents/models. Choose a model whose advertised metadata matches the request, including streaming, tools, JSON output, structured output, reasoning, or image/file inputs. A request that requires an unsupported capability returns the documented API error instead of silently ignoring that requirement.

Use a llmbase_chat_... key when an external agent should consume the user’s chat subscription. Use a llmbase_... inference API key when your application needs direct OpenAI-compatible inference billing, prompt-cache pricing, or the curated inference model list.

Choosing models for agents

Always call GET /api/v1/agents/models for the current model list. Available models can change as the public catalog is updated.

Choose from the returned metadata instead of saving recommendations from this page as a static allowlist:

Need	Metadata to check
Long agent loops	High context_length, low token price, and prompt_cache_supported when available
Tool use	supported_features includes tools
Structured extraction	supported_features includes json_mode or structured_outputs
Reasoning-heavy tasks	supported_features includes reasoning and supported_reasoning_efforts fits your request
Vision or file inputs	input_modalities and media-type metadata match the files your agent sends

For most users, start with a low-cost returned model that supports tools and prompt caching. Move to a larger returned model only when the task needs more context, stronger reasoning, or a capability the smaller model does not expose.