Parameters
The gateway accepts the standard OpenAI chat-completions parameters and passes them through to the upstream model. Support for individual parameters varies by model; unsupported parameters are ignored rather than rejected.
| Parameter | Behavior |
|---|---|
| model | Model slug, e.g. anthropic/claude-sonnet-5. Required. |
| messages | Chat messages array with role and content. Required for chat completions. |
| max_tokens | Upper bound on generated tokens. |
| temperature | Sampling temperature; lower is more deterministic. |
| top_p | Nucleus sampling probability mass. |
| top_k | Top-k sampling, where the model supports it. |
| frequency_penalty | Penalizes tokens by how often they have appeared. |
| presence_penalty | Penalizes tokens that have appeared at all. |
| repetition_penalty | Multiplicative repetition penalty, where supported. |
| stop | One or more sequences that halt generation. |
| stream | Stream tokens as server-sent events when true. |
| tools / tool_choice | Function-calling definitions and selection behavior. |
| response_format | Set { "type": "json_object" } for JSON-mode output, where supported. |
| structured_outputs | Schema-constrained output, where supported. |
| reasoning | Reasoning effort control for reasoning models: effort levels from minimal to xhigh. |
| seed | Best-effort deterministic sampling seed. |