1. Strongest Agentic Coding Model

• State-of-the-art on Terminal-Bench 2.0 (82.7%), Expert-SWE (73.1%), and SWE-Bench Pro (58.6%), outperforming GPT-5.4 on complex coding tasks.
• Holds context across large systems, reasons through ambiguous failures, and carries changes through surrounding codebases with fewer tokens.

2. Higher Intelligence at GPT-5.4 Latency

• Co-designed, trained, and served on NVIDIA GB200/GB300 NVL72 systems to match GPT-5.4 per-token latency while performing at a significantly higher level.
• Uses fewer tokens to complete the same tasks, making it more efficient as well as more capable.

3. Powerful for Knowledge Work & Computer Use

• Scores 84.9% on GDPval (44 occupations) and 78.7% on OSWorld-Verified for autonomous computer operation.
• Excels at generating documents, spreadsheets, and reports; naturally moves across finding information, using tools, and checking output.

4. Scientific Research Co-Scientist

• Leading performance on GeneBench, BixBench, and FrontierMath; helped discover a new proof about Ramsey numbers verified in Lean.
• Strong enough to meaningfully accelerate progress at the frontiers of biomedical and mathematical research.

1. Maximum-compute o-series model

• Uses significantly more compute per query compared to o1.
• Produces deeper, more reliable reasoning chains.
• Best suited for high-stakes tasks that need correctness over speed.

2. Trained with reinforcement learning for deliberate thinking

• Explicit "think-before-answer" architecture.
• Excels at complex reasoning requiring multi-step analysis.

3. Very strong at math, science, coding, and technical proofs

• Handles long derivations, algorithm design, and difficult logic problems.
• Produces structured and explainable reasoning trails.

4. Great for multi-turn reasoning workflows

• Responses API optimized: can think over multiple internal turns before responding.
• Ideal for agentic reasoning pipelines.

5. Large context window

• 200,000-token context for large documents, multi-file review, and long reasoning traces.

6. Multimodal input (text + image)

• Can analyze images for mathematical diagrams, charts, handwritten content, UI layouts, etc.
• Output is text only.

7. Consistency, reliability, and depth

• Designed for situations where accuracy matters more than latency or cost.
• Strong error-checking and self-correction abilities.