OpenAI released its new frontier model, GPT-5.2, on Thursday as competition with Google intensifies. The company positions the model as its most advanced yet, aimed at developers and everyday professionals.

GPT-5.2 arrives in three versions through ChatGPT and the API: Instant, Thinking, and Pro. Each targets different workloads, from fast daily queries to high-stakes reasoning.

During a press briefing, chief product officer Fidji Simo said the company built GPT-5.2 to drive user value.

“We designed 5.2 to unlock even more economic value for people,” she said.

She noted its improvements in spreadsheets, presentations, code, image perception, long-context tasks, and tool use.

The launch comes as Google’s Gemini 3 tops most public benchmarks except coding, where Anthropic still leads.

Competitive pressure rises

GPT-5.2 follows internal urgency at OpenAI. Earlier reports revealed a “code red” memo from CEO Sam Altman after a decline in ChatGPT traffic and concerns that Google is gaining consumer ground.

The memo urged teams to prioritise user experience over new monetisation plans.

This release reflects that shift. OpenAI now aims to reclaim leadership while addressing internal pressure from employees who reportedly wanted more time for refinement.

The company is also doubling down on enterprise use cases despite previous signals that it would focus on consumers.

GPT-5.2 supports that strategy through stronger reasoning, enhanced tool use, and improved long-context handling.

OpenAI says GPT-5.2 sets new benchmark scores in coding, math, science, vision, and long-context reasoning.

The company claims these improvements support “more reliable agentic workflows, production-grade code, and complex systems that operate across large contexts and real-world data.”

Research lead Aidan Clark said the model’s math gains reflect broader progress in logical consistency. “These are all properties that really matter across a wide range of different workloads,” he said.

He added that better math supports financial modeling, forecasting, and advanced data analysis.

OpenAI’s benchmark chart shows GPT-5.2 Thinking outperforming Gemini 3 and Claude Opus 4.5 in most listed reasoning evaluations.

These include real-world software engineering tasks, doctoral-level science tests, and abstract pattern-recognition suites.

Product lead Max Schwarzer said GPT-5.2 delivers notable improvements in code generation and debugging.

“GPT-5.2 makes substantial improvements to code generation and debugging,” he said. He also noted that partner startups report measurable gains in complex multi-step coding workflows.

Schwarzer added that GPT-5.2 Thinking produces 38 percent fewer errors than its predecessor, improving dependability for research, writing, and decision-making.

OpenAI highlights four core upgrades for users: better contextual comprehension, fewer hallucinations, improved STEM reasoning, and sharper creative results.

Free rollout broadens access

OpenAI plans to release GPT-5.2 to all ChatGPT users for free. The company says this move supports its mission to democratize advanced AI.

Everyday users gain smarter assistance for research, planning, writing, and analysis.

Developers receive stronger API tools to build advanced applications.

Educators and creators get more reliable outputs for lessons, ideation, and content production.

OpenAI typically follows major launches with developer tools and transparency updates.

The company expects to release expanded API documentation, updated safety guidelines, and detailed performance breakdowns soon.