Gemini 3.1 Pro

Google llm

Next iteration of the Gemini 3 series, stated by Google to be based on Gemini 3 Pro, with a 1M-token multimodal context and stronger agentic coding.

overall ⌄

69.3

17 of 125 ranked

price · $/M tokens

2 / 12

in / out, as of 2026-07-25

to run it yourself ⌄

API only

no weights to run

Benchmarks

What its maker published, and what anyone else measured. Every figure links the document it came from.

benchmarkpublishedmeasured
GPQA94.3 94.1
SWE-bench80.6 —
SWE-bench Pro54.2 —
HLE44.4 —
BrowseComp85.9 —

Cheaper, and at least as good

On a 1,000-in, 1,000-out call these cost less and rank no lower — a hard comparison to argue with, and a narrow one. Value covers what it misses.

Qwen3.8-Max (Qwen3.8-2.4T-A95B) Alibaba$0.00800overall 69.5
Claude Sonnet 5.5 Anthropic$0.0120overall 72.7

Lineage

↑ finetune Gemini 3 Pro
view in full graph →

Metadata

orgGoogle
released2026-02-19
licenseproprietary