GPT-5.3-Codex Unifies Coding and Reasoning With Near-Double Computer-Use Accuracy

OpenAI

Feb 5, 2026 · Updated Apr 25, 2026

OpenAI released GPT-5.3-Codex, merging GPT-5.2-Codex's frontier coding performance with GPT-5.2's reasoning and professional knowledge into a single model that's 25% faster. It sets new state-of-the-art scores on SWE-Bench Pro and Terminal-Bench while nearly doubling computer-use accuracy on OSWorld-Verified to 64.7%.

OpenAI released GPT-5.3-Codex, combining the coding capabilities of GPT-5.2-Codex with GPT-5.2's broader reasoning into one model. It achieves new highs on SWE-Bench Pro and Terminal-Bench 2.0 while using fewer tokens than any prior model, and scores 64.7% on OSWorld-Verified — up from 38.2% — making it OpenAI's strongest computer-use model for vision-based tasks across Ubuntu, Windows, and macOS.

The model also shares progress and explains decisions as it works, letting you steer tasks in real time instead of waiting for a final output. OpenAI notes it's the first model that was instrumental in creating itself — the team used early versions to debug training and diagnose evaluations. It's also the first model classified as High capability for cybersecurity under OpenAI's Preparedness Framework.

GPT-5.3-Codex is available on all paid ChatGPT plans across the Codex app, CLI, IDE extension, and web, with API access coming soon.

View the full update on openai.com

OpenAI Developers

@OpenAIDevsFeb 5

GPT-5.3-Codex is here. It advances both frontier coding performance and professional knowledge capabilities together in a single model. https://t.co/mAyt3snLY9

181

View on X

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

See all AI news & updates from OpenAI →

Keep reading

OpenAI Upgrades Codex to GPT-5.5 With Expanded Browser and Computer Use

OpenAI updated its Codex agentic coding platform with GPT-5.5, enabling the agent to autonomously navigate web browsers and control desktop applications. This shift allows developers to automate end-to-end workflows like UI testing and asset generation directly within their engineering environment.

GitHubApr 24

GitHub Integrates GPT-5.5 Into Copilot to Solve Complex Agentic Coding Challenges

GitHub launched OpenAI's GPT-5.5 across the Copilot ecosystem, targeting complex, multi-step agentic coding tasks. This rollout signals a shift toward autonomous reasoning models that can resolve real-world challenges previous GPT versions could not handle.

Cursor Adds GPT-5.5 and Leads Internal Benchmarks With 50 Percent Discount

CursorApr 25

Cursor Adds GPT-5.5 and Leads Internal Benchmarks With 50 Percent Discount

Cursor integrated OpenAI's GPT-5.5, which now leads its internal CursorBench evaluation with a 72.8% score. The update highlights a shift toward using real-world developer sessions rather than public benchmarks to measure frontier model performance on complex, multi-file coding tasks.

Windsurf Integrates GPT-5.5 to Power Parallel Agentic Coding Workflows

WindsurfApr 24

Windsurf Integrates GPT-5.5 to Power Parallel Agentic Coding Workflows

Windsurf added OpenAI's new GPT-5.5 model to its AI-native editor to improve how agents handle complex, multi-step engineering tasks. The update focuses on reducing human intervention by better interpreting ambiguous instructions and managing long-horizon goals.