Unverified50% confidenceFactExact time
Under the default configuration, a single failed user request produced 4 upstream model calls, because Spring AI 2.0's OpenAI integration is based on the official openai-java SDK, which has built-in retry settings (spring.ai.openai.chat.max-retries, default 3).
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/19/2026
First Seen
Valid until: 1/17/2027
Sources
Related Entities
Related Claims
VerifiedSpring AI supports multiple model providers including OpenAI, Anthropic, and Ollama, providing a unified API abstraction layer.73% similarUnverifiedThe root of retry multiplication lies in retries occurring at multiple independent layers: inside openai-java SDK there is exponential backoff retry for 429/5xx and connection exceptions (default 3 times), and if application code wraps another retry framework (such as Spring Retry or Resilience4j), a single user click gets amplified into a multiplicative number of upstream calls.70% similarUnverifiedOpenClaw4J基于Java 21和Spring AI构建，为Java开发者提供智能Agent开发框架69% similarUnverifiedThe OpenAI Batch API is a reasonable default when applications are built around OpenAI request shapes and structured output tooling, but applications are responsible for mapping output records back to the frozen list.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1066394API
curl https://kongchang.com/api/v1/knowledge/claims/1066394MCP
get_claim(id=1066394)