Codex GPT-5.6 Sol context window reduced to 258K tokens, users report performance degradation

According to a GitHub issue filed on July 13, Codex GPT-5.6 Sol's context window was reduced from 353K to 258K tokens despite being advertised at 1.05M, with multiple users reporting significant slowness. The issue was opened on the OpenAI Codex repository and labeled a severe regression affecting CLI version 0.144.3 for paid ChatGPT-authenticated users across multiple environments.

Update (2026-07-20): According to GitHub issue tracking and third-party monitoring, OpenAI reduced Codex's maximum context window by 100,000 tokens, from 372k to 272k tokens, while resetting usage limits multiple times. The changes affect active deployments, w...