Google18 Nov 2025Model

Gemini 3 Pro

Made by: Google
Released: 18 Nov 2025
Announced in: Start building with Gemini 3blog.google
Event: Gemini 3 Pro
API record: /api/v1/models/gemini-3-pro

Benchmark results

Benchmark	Score	Source check
Humanity's Last ExamNo tools	37.5%	Checked A person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. Corroborated The same number is printed in 4 other lab documents, the first 3 linked: 1 2 3
Humanity's Last ExamWith tools · search and code execution	45.8%	Checked A person checked this number against its source on 17 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. Corroborated The same number is printed in 5 other lab documents, the first 3 linked: 1 2 3
SWE-bench Verifiedsingle attempt	76.2%	Checked A person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. Corroborated The same number is printed in 12 other lab documents, the first 3 linked: 1 2 3
SWE-bench Prosingle attempt; Thinking (High) · reported later, in Gemini 3.1 Pro: A smarter model for your most complex tasks	43.3%	Checked A person checked this number against its source on 17 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
GPQA Diamondno tools	91.9%	Checked A person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. Corroborated The same number is printed in 10 other lab documents, the first 3 linked: 1 2 3
BrowseCompreported later, in Gemini 3.1 Pro: A smarter model for your most complex tasks	59.2%	Checked A person checked this number against its source on 17 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. Corroborated The same number is printed in 6 other lab documents, the first 3 linked: 1 2 3
ARC-AGI-2ARC Prize Verified	31.1%	Checked A person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. Corroborated The same number is printed in 3 other lab documents: 1 2 3
Terminal-Bench 2.0Terminus 2 harness	54.2%	Checked A person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. Corroborated The same number is printed in 9 other lab documents, the first 3 linked: 1 2 3
Terminus 2 harness; Thinking (High) · reported later, in Gemini 3.1 Pro: A smarter model for your most complex tasks	56.9%	Checked A person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.

Sources

Reported by Google

Start building with Gemini 3blog.google · 18 Nov 2025
Original image
Gemini 3.1 Pro: A smarter model for your most complex tasksblog.google · 19 Feb 2026
Original image

Events

Stories

All models