User Complaint: OpenAI API Rate Limits Throttling Agentic Workflows

Bino5150 · reddit · 2026-09-01

A user reported severe rate limiting issues while testing the OpenAI API with the Luna app. While chat responses were fast, tool calls were heavily throttled, often stopping after just one or two calls with a "try again in 14.753 seconds" error. This forced dozens of manual "continue" turns, bloating context usage and requiring constant babysitting. In contrast, the GLM 5.3 API handled the same task successfully with 54 reasoning steps and 74 cascaded tool calls. The user questioned why a paid provider would enforce such strict throttling.

Original post →

More from coding & agent

coding & agent channel →