> ## Documentation Index
> Fetch the complete documentation index at: https://docs.voidai.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Create a Response

> Use the VoidAI Responses API for supported model input, tools, conversation context, and named streaming events.

`POST https://api.voidai.app/v1/responses`

Requires [authentication](/authentication) and a model that advertises `/v1/responses`. The gateway accepts the OpenAI Responses request format; supported features depend on the selected model/provider.

| Field | Type | Description |
| - | - | - |
| `model` | string, required | Exact model ID. |
| `input` | string or array | Text or Responses-format input items. |
| `instructions` | string | Optional instructions. |
| `stream` | boolean | Request named server-sent events. |
| `max_output_tokens` | number | Optional output limit. |
| `previous_response_id` | string | Provider-dependent continuation context. |
| `tools`, `tool_choice`, `reasoning`, `text` | varies | Model/provider-dependent tools, reasoning, and output-format options. |

```bash theme={null}
curl https://api.voidai.app/v1/responses \
  -H "Authorization: Bearer $VOIDAI_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"model":"gpt-4o-mini","input":"Explain what an API is in one sentence."}'
```

Non-streaming responses contain a response `id`, `output` items, and usage information when available. Streaming uses named events rather than Chat Completions delta chunks. Handle the Responses event format with a compatible client.

A provider may reject an unsupported feature even when the model supports this endpoint. Claude-family requests are also subject to [Claude access and usage checks](/api-reference/messages/create#claude-access-and-usage-checks). See [compaction](/api-reference/responses/compact) and [request usage](/api-reference/usage/get).