SDKv2.5.15
Reduce transient 400 errors in concurrent OpenAI and Azure OpenAI workloads
ContributorAgno Team
April 9, 20261 min read
Resolved an issue where a shared HTTP/2 client was being injected across concurrent OpenAI and Azure OpenAI requests, causing transient 400 errors under load. Each request now uses its own client, eliminating the conflict.
Shipped around the same time
April 9, 2026
Changelog
v2.5.15See audio token usage in run metrics for OpenAI, Perplexity, and LiteLLM
Agno Team
1 min read
April 9, 2026
Changelog
v2.5.15Eliminate duplicate messages from team conversation history
Kaustubh Shukla
1 min read
April 9, 2026
Changelog
v2.5.15Fetch pull requests reliably regardless of repository size
kaiisfree
1 min read