OpenAI Just Released Their Best Model Ever (GPT 5.2)

This hands-on review from The AI Advantage examines GPT 5.2, OpenAI’s newest flagship model, published shortly after its release. The video positions GPT 5.2 as a direct competitive response to Google’s Gemini Deep Think, citing ARC AGI 2 benchmark data from the ARC prize organization: GPT 5.2 Pro scores approximately 38% on medium settings versus Gemini 3 Pro’s 32%, while Gemini Deep Think (the Ultra-tier reasoning model) and GPT 5.2’s top Pro setting sit roughly on par with each other.

Beyond benchmarks, the host runs several practical test prompts. An SVG generation challenge — rendering the Death Star above Los Angeles — provides a visual side-by-side against Gemini Deep Think. The most substantive test involves building a personal budget spreadsheet: both GPT 5.2 and Claude Opus 4.5 are prompted to create a tracker with 10 expense categories, income fields, automated totals, and conditional formatting that turns cells red when budgets are exceeded. Both models produce fully functional spreadsheets with working formulas and formatting, which the host frames as a meaningful milestone — GPT previously lagged Claude significantly on structured spreadsheet tasks.

The overall conclusion is that GPT 5.2 has matched Claude’s spreadsheet capability and reached parity with Gemini’s top reasoning model, which combined with ChatGPT’s existing tooling advantages removes most reasons for users to switch platforms. The video is a useful reference for anyone evaluating frontier model performance on productivity tasks as of late 2025.

📺 Source: The AI Advantage · Published December 12, 2025
🏷️ Format: Benchmark Test