OpenAI has started rolling out GPT 5.1-Codex-Max on Codex with a better performance on coding tasks.
In a post on X, OpenAI confirmed that GPT 5.1-Codex-Max can work independently for hours.
Unlike GPT-5.1, which is optimized for research, normal interaction, generating images, etc, Codex is tailored for coding-related tasks.
Codex works inside the terminal, but you can also use it on the web after connecting it to GitHub.
Over the past few months, Codex has significantly improved, and it competes head-to-head with Claude Code.
In fact, after GPT 5.1, we found Codex better than Claude for complex queries, but Claude Code still wins in adherence.
With GPT-5.1-Codex-Max, OpenAI says Codex is now faster, more capable and token-efficient, and able to work persistently on long tasks with built-in compaction abilities.
“It outperforms GPT-5.1-Codex on real-world engineering tasks and is the first Codex model trained to operate in Windows environments,” OpenAI noted.
“It’s also better at using Powershell, making it a better collaborator on Windows machines.”
On SWE-Bench Verified, it performs better with medium reasoning than GPT-5.1-Codex does while using ~30% fewer thinking tokens.
OpenAI says GPT-5.1-Codex-Max is available today in the Codex CLI, IDE extension, cloud, and code review.
As MCP (Model Context Protocol) becomes the standard for connecting LLMs to tools and data, security teams are moving fast to keep these new services safe.
This free cheat sheet outlines 7 best practices you can start using today.