Codex Repeatedly Compresses and Burns Tokens in Long Tasks
aigclink · x · 2026-07-19
A post pointed out that Codex can insanely burn through tokens in long tasks due to a flaw in its compression design: after compression, the model only sees the initial instructions, causing it to repeatedly enter an "initial instructions -> compress -> execute initial instructions" loop. It might only complete the task by chance if it skips compression; otherwise, it drains the entire quota. The accompanying image further illustrates this runaway loop: once the context is compressed, the working memory is repeatedly reset, leading to redundant execution and uncontrolled token usage. The poster tagged relevant personnel, hoping they would look into the issue.
Related event: Alibaba Announces 2.4T Open-Weight Model Qwen3.8(21 posts)→
More from coding & agent
- Developer uses Replit to run Cursor, Claude and other agents in one codebase — amasad · 2026-07-21
- Antigravity CLI 1.1.5 adds mid-session effort control and model pinning — rseroter · 2026-07-21
- Claude Code is being used to build tiny personal apps when existing tools fall short — alexmacgregor__ · 2026-07-21
- A reply frames AI as a tool for async long-horizon experiments, not just tokens and GPUs — voooooogel · 2026-07-21
- Agent builders argue models should run in micro-VMs instead of tool calling — joecole · 2026-07-21
- Kimi used a forwarded SSH agent and VM API to publish a site on a new machine — davidcrawshaw · 2026-07-21