Unverified50% confidenceFactExact time
The OpenAI Agents SDK's MCPServerStreamableHttp defaults to a 5-second tool call timeout (client_session_timeout_seconds=5).
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/12/2026
First Seen
Valid until: 1/10/2027
Sources
Related Entities
Related Claims
Unverified中心化API（如OpenAI）的首Token延迟（TTFT）通常在200-500毫秒之间73% similarUnverifiedIn this project, loading the model and artifacts from MLflow takes approximately 40 seconds, while the maximum timeout for API Gateway is 30 seconds, making the first invocation almost certain to fail.69% similarUnverifiedMainstream large model API providers typically maintain prompt caches in units of 5 minutes to 1 hour, after which they expire.69% similarUnverifiedAWS SigV4要求客户端与服务器时间偏差不超过5分钟68% similarUnverifiedAfter 2023, end-to-end voice models represented by the OpenAI Realtime API began to emerge, compressing latency from several seconds to under 300 milliseconds.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1008304API
curl https://kongchang.com/api/v1/knowledge/claims/1008304MCP
get_claim(id=1008304)