OpenAI Unveils GPT-5.2 Amid Race for AI Dominance
280OpenAI has introduced GPT-5.2, claiming it delivers major improvements in helping with real-world business tasks. The new version is said to outperform its predecessor, GPT-5.1, which was released last November. It’s now available across different performance tiers—Instant, Thinking, and Pro—offering users more advanced capabilities.
Major Performance Boosts and Benchmark Results
According to OpenAI, GPT-5.2 shows significant progress on various benchmarks. One key test is the GDPval benchmark, which measures how well the model completes 44 different business tasks at the level of human experts. GPT-5.2 matched or exceeded human performance in 70.9% of these tests, compared to GPT-5.1’s 38.8%. These results span the basic Instant tier, the Thinking tier that emphasizes reasoning, and the Pro version designed for research-grade work.
OpenAI highlighted that GPT-5.2 can fully format complex spreadsheets, such as workforce planning tools, while GPT-5.1 could only produce a less refined version. The newer model is better at creating spreadsheets, building presentations, writing code, understanding images, handling long conversations, and managing complex multi-step projects. These improvements aim to unlock more economic value for users across industries.
Performance in Other Key Areas and Real-World Tasks
Beyond GDPval, GPT-5.2 also improves on other benchmarks like ARC-AGI-1 and ARC-AGI-2, which test general problem-solving skills. It also performs better on SWE-Bench Pro and Verified, which involve practical software development tasks. This means the model can more reliably debug code, implement new features, refactor large codebases, and ship fixes with less manual effort.
OpenAI says these enhancements translate into more productive tools for professionals. The new model is expected to reduce the time and effort needed to complete complex technical tasks, making workflows more efficient. GPT-5.2 is now rolling out to ChatGPT users on paid plans, with no change in subscription prices. For API access, it costs $1.75 per million input tokens and $14 per million output tokens, though cached inputs receive a 90% discount. Despite higher per-token costs, OpenAI claims the model’s improved efficiency makes it more cost-effective overall.
Development Speed and Industry Competition
The release of GPT-5.2 shortly after GPT-5.1 signals rapid progress in OpenAI’s development efforts. In early December, CEO Sam Altman sent an emergency memo warning that without swift development of GPT-5, the company risked falling behind Google’s new Gemini 3 model. Since then, OpenAI has indicated that the competitive pressure has eased somewhat. Altman recently told CNBC that Gemini’s advancements were less dramatic than initially feared and that the “code red” alert would end by January.
However, the announcement did not include a direct comparison between GPT-5.2 and Gemini 3’s performance. A separate briefings seemed to suggest only limited comparisons. Industry experts have also questioned how well OpenAI’s benchmarks reflect real-world capabilities, with some analysts noting that benchmark scores don’t always tell the full story of a model’s usefulness in practical scenarios.
Overall, the launch of GPT-5.2 underscores how quickly AI models are evolving and competing. OpenAI aims to maintain its lead by continually improving its models, even as rivals like Google push forward with their own advanced systems. The coming months will likely see further updates and new benchmark tests as the AI race heats up.
What do you think?
It is nice to know your opinion. Leave a comment.