Back in September, OpenAI announced GPT-5-Codex, a new GPT-5 model that is optimized for agentic coding in Codex. Since GPT-5-Codex is based on GPT-5, it comes with significant upgrades in reasoning, coding, and other key areas.

The GPT‑5-Codex model was designed to perform better on real-world software engineering work, including creating new software projects, implementing new features and tests to existing projects, performing large-scale code refactoring tasks, and more.

Recently, OpenAI announced GPT-5-Codex-Mini, a more compact and cost-efficient version of GPT-5-Codex. GPT-5-Codex-Mini will enable developers to enjoy 4x more usage than GPT-5-Codex with a slight capability tradeoff. In the SWE-bench Verified benchmark, GPT-5 High scored 72.8% while the new GPT-5-Codex scored 74.5%. This new GPT-5-Codex-Mini model scores 71.3% on the same SWE-bench Verified benchmark.

OpenAI is recommending developers to select GPT-5-Codex-Mini for easier software engineering tasks and when they are close to hitting rate limits. In fact, Codex itself will suggest developers to switch to GPT-5-Codex-Mini when they hit 90% of their limits. This new mini model is now available in the CLI and IDE extension. API support for this new model is also coming soon.

Thanks to GPU efficiency improvements, OpenAI is now offering a 50% higher rate limit for ChatGPT Plus, Business and Edu users. Also, ChatGPT Pro and Enterprise users will get priority processing for maximum speed.

More predictable usage in Codex. We landed optimizations behind the scenes that mean that you now will get the same usage throughout the day irrespective of the load or how traffic is routed.

Before you could get unlucky and hit a few cache misses in a row and get less usage.

— Tibo (@thsottiaux) November 7, 2025

Finally, the OpenAI team has done some under-the-hood optimizations that will enable developers to enjoy the same Codex usage throughout the day. Previously, based on the load and how the traffic was routed, the usage was affected.