OpenAI Levels Up Coding Assistance with GPT-5.1-Codex-Max

Developers got an early holiday present this week as OpenAI launched GPT-5.1-Codex-Max, a souped-up version of its programming assistant that delivers significant improvements while maintaining the same price point as its predecessor.

What's New Under the Hood?

The standout feature? This model can run continuously for over 24 hours thanks to advanced "compaction" technology that handles context windows exceeding one million tokens. Imagine having a coding partner that doesn't need coffee breaks - that's essentially what OpenAI has created here.

Benchmark tests tell an impressive story:

• 77.9% on SWE-Bench Verified
• 79.9% on SWE-Lancer IC SWE
• 58.1% on TerminalBench2.0

These numbers represent clear jumps over the standard GPT-5.1-Codex, particularly when tackling multi-step development challenges.

Windows Developers Rejoice

In a strategic shift, OpenAI specifically optimized this model for Windows environments during training - a welcome change from previous Unix-focused versions. "We're seeing more developers working in Windows ecosystems," explains an OpenAI spokesperson, "and we wanted our tools to meet them where they work."

The pricing strategy makes this upgrade even sweeter:

• Input: $1.25 per million tokens
• Output: $10 per million tokens

Where You'll Find It

The new model is already powering up several developer favorites:

• Cursor
• GitHub Copilot (available to Pro, Pro+, Business and Enterprise users)
• Linear

For teams working on complex projects requiring sustained coding sessions, this could be the productivity boost they've been waiting for - all without the dreaded price hike that often accompanies tech upgrades.

Key Points:

• Extended runtime: 24+ hours of continuous operation
• Windows optimization: First Codex model built for Windows-first developers
• Price parity: Same cost as previous version despite performance gains
• Widespread integration: Available now in major coding platforms