# Reduce transient 400 errors in concurrent OpenAI and Azure OpenAI workloads

> 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.

- Published: 2026-04-09
- Author: Agno Team
- Category: Changelog
- Canonical: https://agno-com-nine.vercel.app/articles/reduce-transient-400-errors-in-concurrent-openai-and-azure-openai-workloads
- Markdown: https://agno-com-nine.vercel.app/articles/reduce-transient-400-errors-in-concurrent-openai-and-azure-openai-workloads.md

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.