OpenAI has unveiled its newest AI model, GPT-5.2. This model is being contrasted with Google’s Gemini 3, its primary rival.
The creator of ChatGPT launched GPT-5.2 on Thursday, initially for paid ChatGPT subscribers, with plans for wider access to follow. This release comes after a leaked memo in which OpenAI CEO Sam Altman declared “code red” owing to heightened competition from Google’s Gemini 3. Google’s newest AI model has been surpassing its rivals in various benchmark assessments and on the AI rankings at LMArena.
Moreover, a post by former Google employee Deedy Das highlighted that OpenAI experienced a nearly 6 percent drop in its visitors within two weeks following the debut of Gemini 3.
How to experiment with GPT-5.2, the latest series of ChatGPT models from OpenAI
How does GPT-5.2 stack up against Gemini 3? At present, both models are closely matched in benchmark performance and pricing, each showcasing strengths in particular domains.
It’s noteworthy that GPT-5.2 has just premiered, and further details will be revealed shortly. OpenAI asserts that GPT-5.2 shines in “professional knowledge tasks,” encompassing spreadsheet creation, presentation construction, coding, image perception, long context comprehension, tool utilization, and managing intricate projects.
GPT-5.2 vs. Gemini 3: LMArena rankings
Based on the LMArena leaderboards, GPT-5.2 is excelling in web development. The GPT-5.2-high model occupies the second spot, trailing Claude Opus 4.5, while Gemini 3 Pro sits at fourth, and the base GPT-5.2 model is ranked sixth. These represent preliminary outcomes.
GPT-5.2 has not yet achieved a ranking on the other LMArena leaderboards.
However, GPT-5.2 encounters obstacles. Different Gemini 3 variants currently lead in text, vision, text-to-image, image editing, and search categories, with Google Veo 3 models ahead in text-to-video and image-to-video classifications. Additionally, Gemini 3 leads the overall leaderboard, where GPT-5.2 remains unranked.
GPT-5.1-high currently holds sixth place in the overall standings, possibly clarifying Altman’s uneasiness.
GPT-5.2 vs. Gemini 3: Benchmark assessments
Independent validation of GPT-5.2’s benchmark outcomes is not yet accessible, and <a href="https://scale.com/leaderboard/swe_bench_pro_public" target="_blank" data-ga-click="1" data-ga-label="$text" data-ga-item="text-link" data-ga