GPT-Realtime-2
语音到语音的实时语音模型，推理强度可配置，工具使用可靠。
规格
- 上下文
- 128K
- 最大输出
- 32K
- 输入价
- —
- 输出价
- —
- 发布日期
- 2026-05-07
上下文
- 上下文
- 128K
- 最大输入
- —
- 最大输出
- 32K
价格
- 官方未公布
输入输出
- 输入
- 文本音频图像
- 输出
- 文本音频
API
- 接口类型
思考
- 支持思考
- 支持
- 思考等级
- —
源信息
- 状态
- 可用
- 发布日期
- 2026-05-07
- 知识截止
- 2024-09-30
能力
- 已确认
调用方式
1 个 Provider
openaimodel = gpt-realtime-2
realtime
GEThttps://api.openai.com/v1/realtime
JSON
标准格式
{
"id": "gpt-realtime-2",
"object": "model",
"created": 1778112000,
"owned_by": "openai",
"name": "GPT-Realtime-2",
"api": {
"types": [
"realtime"
]
},
"limits": {
"context": 128000,
"input": null,
"output": 32000
},
"modalities": {
"input": [
"text",
"audio",
"image"
],
"output": [
"text",
"audio"
]
},
"reasoning": {
"supported": true,
"efforts": []
},
"pricing": null,
"features": [
"tools",
"caching"
],
"info": {
"status": "active",
"release_date": "2026-05-07",
"knowledge_cutoff": "2024-09-30",
"description": "Speech-to-speech realtime voice model with configurable reasoning effort and reliable tool use.",
"docs": "https://developers.openai.com/api/docs/models/gpt-realtime-2",
"verified_at": "2026-10-11"
}
}官方来源
4