OpenAI just released GPT-5.5 on April 23, 2026—the smartest and most intuitive model yet, designed for real work on a computer.
The Benchmark Numbers
GPT-5.5 crushes previous models across key benchmarks:
| Benchmark | GPT-5.5 | GPT-5.4 | Claude Opus 4.7 | Gemini 3.1 Pro |
|---|---|---|---|---|
| Terminal-Bench 2.0 | 82.7% | 75.1% | 69.4% | 68.5% |
| Expert-SWE | 73.1% | 68.5% | - | - |
| GDPval (wins/ties) | 84.9% | 83.0% | 80.3% | 67.3% |
| OSWorld-Verified | 78.7% | 75.0% | 78.0% | - |
| FrontierMath Tier 1-3 | 51.7% | 47.6% | 43.8% | 36.9% |
GPT-5.5 Pro hits 90.1% on BrowseComp (vs 84.4% for standard GPT-5.5).
What Makes It Different
GPT-5.5 matches GPT-5.4 per-token latency while performing at much higher intelligence. Larger models usually mean slower serving—GPT-5.5 breaks that tradeoff.
It uses significantly fewer tokens to complete the same Codex tasks. More capable AND more efficient.
The Agentic Leap
The gains are strongest in:
- Agentic coding
- Computer use
- Knowledge work
- Early scientific research
Give GPT-5.5 a messy, multi-part task and it plans, uses tools, checks its work, navigates ambiguity, and keeps going. No micromanaging every step.
Availability
Rolling out now to Plus, Pro, Business, and Enterprise users in ChatGPT and Codex. API coming "very soon."
Safety
Strongest safeguards to date. Evaluated across full safety frameworks, tested by internal and external redteamers, nearly 200 trusted early-access partners reviewed real use cases.