Model Field

Provider

LLM service provider for this model slot.

— select —

Model name

Model identifier. Click the field or search icon to browse available models.

No models found

API key

Authentication key for this provider. Shared across all model slots using the same provider.

{ if (v) $store.modelConfig.apiKeyValues[prov] = v; }); } " :name="showKey ? 'visibility' : 'visibility_off'">

API key

Authentication key for this provider. Shared across all model slots using the same provider.

{ if (v) { model.api_key = v; _revealed = v; } }); } if (!showKey && _revealed && model.api_key === _revealed) { model.api_key = ''; _revealed = ''; } " :name="showKey ? 'visibility' : 'visibility_off'">

Supports Vision

Models capable of Vision can for example natively see the content of image attachments.

Use separate Vision Model

When disabled, vision_load uses this model's native vision.

Context window size

Maximum number of tokens in the context window. System prompt, chat history, RAG and response all count towards this limit.

Advanced Settings

API base URL

Custom endpoint URL. Leave empty to use the provider's default.

Context window space for chat history

Portion of context window dedicated to chat history visible to the agent. Smaller size will result in shorter and more summarized history.

Max embeds

Maximum number of embedded images used by the chat model. Set to 0 for unlimited.

Max embeds

Maximum number of images sent to the Vision Model in one call. Set to 0 for unlimited.

Timeout (seconds)

How long to wait for the Vision Model.

Max tokens

Maximum output tokens for each delegated vision call.

To customize the vision prompt, use Agent Editor > Advanced > Prompt files and edit fw.vision_load.md.

Context window space for utility model input

Portion of context window used for utility model input messages.

Requests per minute limit

Limits the number of requests per minute. Waits if the limit is exceeded. Set to 0 to disable.

Input tokens per minute limit

Limits the number of input tokens per minute. Waits if the limit is exceeded. Set to 0 to disable.

Output tokens per minute limit

Limits the number of output tokens per minute. Waits if the limit is exceeded. Set to 0 to disable.

Additional parameters

Any other parameters supported by LiteLLM. Format is KEY=VALUE on individual lines. Value can be JSON objects; unquoted is treated as object/number, quoted as string.