OpenAI has officially launched GPT-5.2 this Thursday, currently available to paid ChatGPT users, with broader access expected in the coming weeks. The launch coincides with reports of internal tension at OpenAI, following a leaked memo where CEO Sam Altman declared a “code red” in response to increasing competition from Google’s latest AI model, Gemini 3.
Rising Rivalry: OpenAI vs Google
Google’s Gemini 3 has been making waves as the tech giant’s “most intelligent” AI model to date. It has surpassed competitors in multiple benchmark tests and secured top positions on LMArena’s AI leaderboards. In addition, a post from former Google employee Deedy Das indicated that OpenAI lost nearly 6% of its visitors within two weeks of Gemini 3’s release, highlighting the intensifying rivalry between the two AI powerhouses.
Comparing GPT-5.2 and Gemini 3
Benchmark Performance
OpenAI claims that GPT-5.2 excels in professional knowledge work, such as creating spreadsheets, building presentations, coding, interpreting images, managing long contexts, and executing complex multi-step tasks. Despite these claims, benchmark comparisons show a mixed picture:
- SWE-bench Verified: GPT-5.2 – 80%, Gemini 3 – 76.2%
- Humanity’s Last Exam (HLE, no tools): GPT-5.2 – 34.5%, Gemini 3 – 37.5%
- GPQA Diamond: GPT-5.2 – 92.4%, Gemini 3 – 91.9%
- AIME 2025 (no tools): GPT-5.2 – 100%, Gemini 3 – 95%
- MMMLU: GPT-5.2 – 89.6%, Gemini 3 – 91.8%
Both models win in certain areas and fall behind in others, making it a close contest for now.
LMArena Leaderboard Insights
Early leaderboard results suggest GPT-5.2 performs particularly well in web development. The higher-end GPT-5.2-high model ranks second, just behind Claude Opus 4.5, while Gemini 3 Pro sits in fourth place. The base GPT-5.2 model ranks sixth. However, Gemini 3 dominates the overall leaderboard across categories such as text, vision, text-to-image, and search, with related Google Veo 3 models topping video-related categories.
Features and Platform Integration
Both GPT-5.2 and Gemini 3 are available as series of models integrated into their respective platforms—ChatGPT and Google Gemini. Key differences include:
- Gemini 3: Generates AI images and videos, integrated across Google apps and Google AI Studio tools.
- GPT-5.2: Generates AI images natively, while video creation requires the Sora app.
Google’s broader integration provides users with more access points and versatility, although GPT-5.2 remains highly capable in specialized tasks.
Pricing Comparison
Both AI models are competitively priced for subscription and API access:
- ChatGPT Plus (GPT-5.2): $20/month; Pro plan $200/month
- Google AI Pro (Gemini 3): $20/month; Ultra plan $249.99/month
- API Pricing:
- GPT-5.2 – $1.75 per 1M input tokens, $14 per 1M output tokens
- Gemini 3 – $2 per 1M input tokens, $12 per 1M output tokens
The pricing remains close, giving users flexibility to choose based on performance needs rather than cost alone.
Which AI Model Should You Choose?
Given the early stages of GPT-5.2’s release and the dominance of Gemini 3 in several leaderboard categories, the best choice depends on your use case. Users seeking advanced professional productivity tools may prefer GPT-5.2, while those needing broader platform integration and multimedia capabilities may lean toward Gemini 3. The most effective approach is to test both models and determine which aligns best with your workflow and goals.
About the Author
