OpenAI Chat Completions(HTTP)¶
OpenClaw 的 Gateway 可以提供一個小型、與 OpenAI 相容的 Chat Completions 端點。
此端點預設為停用。請先在設定中啟用。 請先在設定中啟用。
POST /v1/chat/completions- 與 Gateway 使用相同的連接埠(WS + HTTP 多工):
http://<gateway-host>:<port>/v1/chat/completions
在內部實作上,請求會以一般的 Gateway 代理程式執行流程來處理(與 openclaw agent 相同的程式碼路徑),因此路由、權限與設定都會符合你的 Gateway。
驗證¶
Uses the Gateway auth configuration. 傳送 Bearer 權杖:
Authorization: Bearer <token>
注意事項:
- 當
gateway.auth.mode="token"時,請使用gateway.auth.token(或OPENCLAW_GATEWAY_TOKEN)。 - 當
gateway.auth.mode="password"時,請使用gateway.auth.password(或OPENCLAW_GATEWAY_PASSWORD)。
選擇代理程式¶
不需要自訂標頭:在 OpenAI 的 model 欄位中編碼代理程式 ID:
model: "openclaw:<agentId>"(例如:"openclaw:main"、"openclaw:beta")model: "agent:<agentId>"(別名)
或透過標頭指定特定的 OpenClaw 代理程式:
x-openclaw-agent-id: <agentId>(預設:main)
進階:
x-openclaw-session-key: <sessionKey>以完整控制工作階段路由。
Enabling the endpoint¶
將 gateway.http.endpoints.chatCompletions.enabled 設為 true:
{
gateway: {
http: {
endpoints: {
chatCompletions: { enabled: true },
},
},
},
}
停用端點¶
將 gateway.http.endpoints.chatCompletions.enabled 設為 false:
{
gateway: {
http: {
endpoints: {
chatCompletions: { enabled: false },
},
},
},
}
Session behavior¶
By default the endpoint is stateless per request (a new session key is generated each call).
If the request includes an OpenAI user string, the Gateway derives a stable session key from it, so repeated calls can share an agent session.
串流(SSE)¶
將 stream: true 設定為接收 Server-Sent Events(SSE):
Content-Type: text/event-stream- 每一行事件為
data: <json> - 串流以
data: [DONE]結束
範例¶
非串流:
curl -sS http://127.0.0.1:18789/v1/chat/completions \
-H 'Authorization: Bearer YOUR_TOKEN' \
-H 'Content-Type: application/json' \
-H 'x-openclaw-agent-id: main' \
-d '{
"model": "openclaw",
"messages": [{"role":"user","content":"hi"}]
}'
串流:
curl -N http://127.0.0.1:18789/v1/chat/completions \
-H 'Authorization: Bearer YOUR_TOKEN' \
-H 'Content-Type: application/json' \
-H 'x-openclaw-agent-id: main' \
-d '{
"model": "openclaw",
"stream": true,
"messages": [{"role":"user","content":"hi"}]
}'