OpenAI Codex Infinite Loop Bug Suspected of Doubling Token Usage
nlight · reddit · 2026-08-05
Heavy users report that OpenAI Codex has been experiencing a severe infinite loop bug since the July 22 update. Even with clear goals, the agent loops for days, especially in ultra thinking modes.
Analysis reveals that this issue leads to a massive increase in token usage over time (suspected doubling). The developer's investigation suggests it might be related to changes in version 0.146 and the waitagentenabled config setting, with multiple related issues already tracked on GitHub.
More from coding & agent
- Weaviate Query Agent Introduces Suggest Queries Mode — CShorten30 · 2026-08-05
- Developer Proposes Building the First AI Agent Harness Benchmark — Comfortable-Rock-498 · 2026-08-05
- Codex Integrates GlobalGPT via MCP for Multimodal Generation — FellMentKE · 2026-08-05
- Mixed-Precision Quantization for vLLM: Qwen3.6 27B Models at 2-8bit — victormustar · 2026-08-05
- Developers seek open-source coding models as cost-effective alternatives to Claude and Codex — lakshya_sharma_3 · 2026-08-05
- French Indie Hacker Sells AI Content Tool for €3,000 Using 19 AI Agents — huangyun_122 · 2026-08-05