Gemini 3 Pro

Gemini 3 Pro, released on 18 November 2025, is the model that opened Google's third Gemini generation. It takes text, images, audio and video on input and answers in text, with a context window of one million tokens and output capped at roughly 65,500. Its knowledge cutoff is January 2025. At launch it took the lead on several public benchmarks: 95% on AIME 2025, 91.9% on GPQA Diamond, 76.2% on SWE-bench Verified and 31.1% on ARC-AGI-2, together with the top position on the LMArena leaderboard at around 1501 Elo. The jump on ARC-AGI-2, a test built specifically to resist memorisation, was the result the field paid most attention to, and Google paired the model with a separate Deep Think mode for problems that need longer deliberation. Pricing at release was two dollars per million input tokens and twelve per million output. The model remains the reference point for everything that followed in the line: the later Gemini 3.1 Pro roughly doubled the ARC-AGI-2 score, and the lightweight Gemini 3.6 Flash took the same generation into cheap, high-volume serving. Both have their own entries in this catalogue.

▸Videos

No videos yet.

▸Reviews