The base URL for the Openai Compatible Inference API.
The API Key for Authentication.
Fixed delay in milliseconds between consecutive requests.
Number of lines to read when using `read_file` tool. Large lines may cost more tokens. Default is 0, let
model decide lines.
Retry Configuration
Enable retry mechanism for api errors.
Maximum number of retry attempts.
Interval between retry attempts in milliseconds.
Additional HTTP status codes that will be merged.
Git Commit Message
Select the model to be used for Git commit message generation.
Language for generated Git commit messages.
Copilot Compatibility
Disabled by default. When enabled, this exposes a hidden gpt-4o-mini alias backed by the selected
model; use only for Copilot Agent network failures because it may affect reasoning UI.
Provider Management
Provider ID
Base URL
API Key
API Mode
Custom Headers (JSON)
Actions
Model Management
Model ID
Provider ID
Display Name
Config ID
Context Length
Max Tokens
Supports Vision
Temperature
TopP
Delay (ms)
Actions
Add New Model
Model provider.
Model support vision.
'openai' for API (/chat/completions), 'openai-responses' for API (/responses), 'ollama' for API
(/api/chat), 'anthropic' for API (/v1/messages), 'gemini' for API
(/v1beta/models/{model}:streamGenerateContent?alt=sse).
Model ID (e.g., gpt-4, claude-3).
Select Model
Configuration ID for this model. Allows defining the same model with different settings (e.g.
'glm-4.6::thinking', 'glm-4.6::no-thinking').
Base URL for the model provider.
Maximum context length.
Maximum number of tokens to generate (range: [1, context_length)).
Maximum output tokens (OpenAI new standard - takes precedence over Max Tokens if both are set).
Sampling temperature (range: [0, 2]). Default is 0.
Top-p sampling value (range: (0, 1]).
Model-specific delay in milliseconds between consecutive requests.
Display name for the model that will be shown in the host interface.
Model family (e.g., 'gpt-4', 'claude-3', 'gemini'). Enables model-specific optimizations and
behaviors.
Top-k sampling value (range: [1, Infinity)).
Minimum probability threshold (range: [0, 1]).
Maximum number of tokens for chain-of-thought output.
Frequency penalty (range: [-2, 2]).
Presence penalty (range: [-2, 2]).
Repetition penalty (range: (0, 2]).
Include reasoning_content in assistant messages sent to the API.
Include "thinking.type" in request body.
Include "enable_thinking" in request body.
Default OpenAI reasoning effort. Leave empty to hide the host picker.
Reasoning Configuration (OpenRouter)
Enable reasoning params in request body.
Include "reasoning.effort" in request body.
Include "reasoning.exclude" in request body.
Include "reasoning.max_tokens" in request body.
Custom HTTP headers to be sent with every request to this model.
Extra request body parameters to be sent with every request to this model.