News
Breaking: OpenAI Launches GPT-5.2, Its Most Capable Model Yet for Professional Work
OpenAI launched GPT-5.2 — Instant, Thinking, and Pro variants rolling out to paid ChatGPT users first, with big benchmark gains over GPT-5.1.

OpenAI has launched GPT-5.2, its new flagship model series, rolling out first to paid ChatGPT users. The company calls it the most capable model series yet for professional knowledge work. The launch lands days after CEO Sam Altman declared an internal "code red" to refocus the company on ChatGPT following Google's Gemini 3 Pro release. GPT-5.2 ships in three variants — Instant, Thinking, and Pro — and OpenAI says it is faster, more accurate, and far better at writing, coding, and long tasks than GPT-5.1.
What we know
- Three variants, one series. Instant is the fast option for everyday lookups; Thinking handles coding, math, and planning; Pro is the most powerful tier for the hardest questions. All three are rolling out to ChatGPT Plus, Team, and Enterprise subscribers first, and to developers through the API on paid tiers.
- Big benchmark jumps, on OpenAI's numbers. On GDPval, which compares models with human professionals across 44 occupations, GPT-5.2 Thinking met or exceeded human expert performance 70.9% of the time — up from 38.8% for GPT-5 and ahead of Claude Opus 4.5 (59.6%) and Gemini 3 Pro (53.3%). On the SWE-Bench Pro coding test it scored 55.6%, about 5 points better than GPT-5.1.
- A much bigger working memory. GPT-5.2 carries a 400,000-token context window with up to 128,000 tokens of output, for entire codebases or long documents in one request.
- Fewer hallucinations, OpenAI claims. The company says GPT-5.2 Thinking makes about 30% fewer factual errors than GPT-5.1, and it plans to phase GPT-5.1 out after a transition of roughly three months. API pricing is $1.75 per million input tokens and $14 per million output tokens.
- The "code red" context. Altman ordered the company-wide push after Gemini 3 Pro topped industry leaderboards. Fidji Simo, OpenAI's CEO of applications, told reporters the model had been in development for months and was not rushed out — it was known internally by the code name "Garlic."
What's unconfirmed
- Free-tier timing. OpenAI has not announced when GPT-5.2 reaches free ChatGPT users. Paid plans have it first; everyone else waits.
- Independent verification. The benchmark figures above are OpenAI's own. No major independent index has scored GPT-5.2 yet, so treat the performance claims as the company's until third-party tests land.
- How the rollout actually feels. A phased rollout means some paying users will see it before others, and real-world reliability will only show up in everyday use.
Why it matters
For paid ChatGPT users, GPT-5.2 is a genuine upgrade: stronger coding and writing, a far larger context window, and fewer wrong answers, at no change to plan prices. For everyone else, the near-term effect is narrow — free users keep GPT-5.1 for now, and the real test is whether the benchmark gains survive contact with real work. The bigger signal is the race itself: Google's Gemini 3 forced OpenAI into a public sprint, and the two labs are now trading flagship launches weeks apart.
Sources: Red Hot Singapore, TechPowerUp, Creati.ai. Benchmark figures are OpenAI's own unless noted.