---
title: "Chat-completions style inference (OpenAI-compatible shape for some models)"
method: POST
path: "/chat/completions"
tags: ["Chat"]
---

# Chat-completions style inference (OpenAI-compatible shape for some models)

`POST /chat/completions`

## Request body

- CreateChatCompletionRequest
  - `model` string, required — Model identifier from the ZeroGPU dashboard.
  - `messages` ChatMessage[], required
    - `role` 'system' | 'user' | 'assistant', required — Message role
    - `content` string, required — Message text
  - `metadata` object — Optional model-specific parameters (e.g. PII options).

## Response `200`

Success

- ChatCompletionResponse — Chat completion payload (shape varies by model; extra fields allowed).
  - `id` string
  - `object` string
  - `created` integer
  - `model` string
  - `choices` object[]
  - `usage` TokenUsage
    - `input_tokens` integer
    - `output_tokens` integer
    - `total_tokens` integer

## Other responses

- `400` — Bad request — invalid body
- `401` — Unauthorized — invalid or missing API key
- `403` — Forbidden — invalid project ID or permissions
- `420` — Input exceeds model token limit (context_length_exceeded)
- `500` — Internal server error

## Changes

- **2026-05-12** `7d52cb055bba` — 1 info
  - endpoint added

[Change history](https://skmtc.dev/zerogpu/apis/zerogpu-api/changes/chat/completions/post.md)

---

[API](https://skmtc.dev/zerogpu/apis/zerogpu-api.md) · [All operations](https://skmtc.dev/zerogpu/apis/zerogpu-api/llms.txt) · [OpenAPI document](https://skmtc-service-production.skmtc.workers.dev/v1/apis/zerogpu/zerogpu-api/revisions/7d52cb055bba/schema)