GPT‑5.2 codename Robin is out but I wouldn't trust the official benchmarks. According to Lynchmark, GPT‑5.2 does not outperform Gemini 3 Pro just yet.
Look at the full benchmark results to compare for yourself. The data shows Gemini 3 Pro still leads in coding accuracy across a range of real‑world browser‑based JavaScript tasks.
While GPT‑5.2 shows improvement over previous OpenAI models, it hasn't closed the gap with Google's top offering. The race is closer, but the crown hasn't changed hands.
Don't take marketing claims at face value—check the numbers. Lynchmark runs automated, reproducible tests that measure actual implementation ability, not just theoretical performance.