# isReasoningModel

**Kind:** Function

**Source:** [`atloria-monorepo/packages/doc-automation/src/v2/providers/azure-responses.provider.ts`](https://github.com/sherkety/atloria/blob/main/atloria-monorepo/packages/doc-automation/src/v2/providers/azure-responses.provider.ts#L47)

GPT-5 series (gpt-5, gpt-5.1/.2/.4/.5, gpt-5.3-codex) and the o-series (o1/o3/o4)
are REASONING models. On the Responses API they reject sampling params
(`temperature`, `top_p`, penalties, `stop`, `max_tokens`) with a 400, and are
instead steered by `reasoning.effort` + `text.verbosity`. The non-reasoning chat
variant `gpt-5-chat` still takes `temperature`, so it's excluded here.

## Signature

```ts
function isReasoningModel(model: string): boolean
```

## Parameters

| Name | Type |
|---|---|
| `model` | `string` |

**Returns:** `boolean`