gpt-5-pro-official API 模型解析指南
快速结论
- Supports Chat Completions for low-friction SDK migration.
- Supports streaming and tool-calling for agent workflows.
- Stabilize system/messages parameters before production rollout.
核心能力
- Chat and reasoning：Supports multi-turn context and structured outputs.
- Streaming output：SSE token streaming for real-time UX.
- Tool calling：Compose retrieval/execution workflows with tools.
适用场景
- When building coding/Q&A assistants and task agents.
- When streaming and tool-calling are required.
不适用场景
- For pure image/video generation tasks.
- For simple templated text without reasoning needs.
运行特性
- API calls are stateless; pass full context via messages.
- With stream=true, responses return as SSE events.
最小请求示例
{
"model": "gpt-5-pro-official",
"messages": [
{
"role": "user",
"content": "Summarize suitable use cases for this model."
}
],
"max_tokens": 200
}
最小响应示例
{
"id": "chatcmpl_xxxxxxxx",
"model": "gpt-5-pro-official",
"object": "chat.completion",
"choices": [
{
"index": 0,
"finish_reason": "stop"
}
]
}
关键参数
| 参数 | 类型 | 必填 | 默认值 | 范围 | 说明 |
|---|
| model | string | 是 | gpt-5-pro-official | - | Exact model identifier from ToAPIs. |
| messages | object[] | 是 | - | - | Conversation messages in chronological order. |
| max_tokens | integer | 是 | - | >=1 | Maximum output tokens. |
| stream | boolean | 否 | false | - | Enable SSE streaming output. |
| tools | object[] | 否 | - | - | Tool/function definitions for tool-calling. |
常见错误
| HTTP | Code | 触发条件 | 修复建议 | 重试策略 |
|---|
| 400 | invalid_request_error | Missing fields or invalid payload types. | Validate model/messages/max_tokens schema. | Retry only after payload fix. |
| 401 | authentication_error | Missing or invalid API key. | Verify Authorization header and key scope. | Retry after authentication fix. |
FAQ
- What is gpt-5-pro-official best for?
Best for chat, reasoning, coding, and tool-calling assistant workflows.
相关 API
