feat(fam-edge): 云端模型调用统计 - 新增 model_calls 表(provider/model/时间/耗时/成功/失败原因),gemini/nvidia 每次请求经 model_call_hook 记录并随 sync delta 下发;gemini 支持模型级 model_timeouts(lite 实测22.6s×4≈90s)
This commit is contained in:
@@ -55,6 +55,10 @@ models:
|
||||
- "gemini-flash-lite-latest"
|
||||
api_key: "${GEMINI_API_KEY}"
|
||||
timeout: 600
|
||||
# 模型级独立超时(最终值,不参与编排层 ×2 放大)
|
||||
# gemini-flash-lite 实测耗时 ~22.6s(真实 370MB 监控视频),按用户要求 ×4 ≈ 90s
|
||||
model_timeouts:
|
||||
"gemini-flash-lite-latest": 90
|
||||
circuit_breaker:
|
||||
enabled: true
|
||||
threshold: 5
|
||||
|
||||
Reference in New Issue
Block a user