Parameter support varies by endpoint and model. Use this page as a quick reference, then confirm details on each endpoint page.
Common request parameters
| Parameter | Type | Typical usage |
|---|
model | string | Required model id (for example openai/gpt-5-nano). |
stream | boolean | Enable SSE token streaming on text endpoints. |
temperature | number | Control output randomness when supported. |
top_p | number | Nucleus sampling control where supported. |
max_tokens / max_output_tokens | integer | Cap output length and cost. |
stop | string or string[] | Define explicit stop sequences. |
tools, tool_choice | object / string | Tool-calling controls on compatible endpoints. |
response_format | string or object | Ask for plain text, JSON, or schema-constrained responses. |
meta, usage | boolean | Include extra metadata and usage details in responses. |
Provider routing parameters
The provider object lets you influence routing behavior:
| Field | Type | Purpose |
|---|
order | string[] | Preferred provider order. |
only | string[] | Restrict routing to specific providers. |
ignore | string[] | Exclude specific providers. |
include_alpha | boolean | Allow alpha providers in routing decisions. |
Provider-specific options
Use provider_options when you need provider-native configuration (for example cache controls). Keep these isolated so you can preserve portability between providers.
Debug parameters
Use debug for controlled troubleshooting:
| Field | Type | Purpose |
|---|
enabled | boolean | Enable debug mode for the request. |
return_upstream_request | boolean | Include transformed upstream request payload. |
return_upstream_response | boolean | Include upstream response payload where available. |
trace | boolean | Return routing/debug traces. |
trace_level | summary or full | Control trace verbosity. |
debug data may contain sensitive request context. Use only in development or controlled environments.
Example
{
"model": "openai/gpt-5-nano",
"input": "Summarize this changelog.",
"stream": false,
"temperature": 0.3,
"max_output_tokens": 300,
"provider": {
"order": ["openai", "anthropic"],
"ignore": ["some-provider"]
},
"debug": {
"enabled": true,
"trace": true,
"trace_level": "summary"
}
}
Related pages
Last modified on April 21, 2026
