Google's currently-shipping frontier reasoning model; the multimodal breadth leader.
Gemini 3.1 Pro is Google's currently-shipping frontier model and the survey's breadth leader on modality: it handles text, image, audio, and video natively, a wider input surface than any other flagship here. At $2 / $12 per million tokens it is also the most aggressively priced model in the flagship tier.
What it lacks, as of the August survey, is a current independent intelligence score — so its exact placement against Opus 5, Fable 5, and Sol is unresolved here, and we flag that rather than lean on Google's own numbers. A newer Gemini 3.5 Pro, rumored to carry a 2M-token context, remains unreleased, which leaves 3.1 Pro as the model to actually build on today.
For multimodal applications — anything spanning audio or video, or needing several input types in one model — Gemini 3.1 Pro is the strongest and cheapest flagship option. For pure text reasoning at the very top, the independently ranked closed models currently have the clearer claim. As with the rest of the flagship tier, the weights are closed, so self-hosting is not available.
FIELD NOTE — The newer Gemini 3.5 Pro (rumored 2M context) remains unreleased as of August.
VISIT OFFICIAL SITE →