GPQA Diamond Effort: High 91.9% blog.google for Gemini 3 Pro (high) on Graduate-Level Google-Proof Q&A Diamond (opens in a new tab) GPQA Diamond Accuracy Reported November 18, 2025 Launch post Google Self-reported Google DeepMind Record for Gemini 3 Pro on Graduate-Level Google-Proof Q&A Diamond GPQA Diamond — Epoch AI harness Configuration: Epoch AI harness 92.6% epoch.ai for Gemini 3 Pro on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness (opens in a new tab) Archived copy (opens in a new tab) GPQA Diamond Accuracy Reported October 7, 2026 Evaluated November 19, 2025 at 3:16 PM UTC Benchmark leaderboard Epoch AI Independent Record for Gemini 3 Pro on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness GPQA Diamond — Vals AI harness Effort: High Configuration: Vals AI harness: mean of zero-shot and few-shot CoT 91.7% vals.ai for Gemini 3 Pro (high) on Graduate-Level Google-Proof Q&A Diamond — Vals AI harness (opens in a new tab) Archived copy (opens in a new tab) GPQA Diamond Accuracy Reported September 1, 2026 Benchmark leaderboard Vals AI Independent Record for Gemini 3 Pro on Graduate-Level Google-Proof Q&A Diamond — Vals AI harness Humanity's Last Exam Full set — No tools Effort: High Configuration: No tools 37.5% blog.google for Gemini 3 Pro (high) on Humanity's Last Exam Full set — No tools (opens in a new tab) Humanity's Last Exam Accuracy Reported November 18, 2025 Launch post Google Self-reported Google DeepMind Record for Gemini 3 Pro on Humanity's Last Exam Full set — No tools Humanity's Last Exam Full set — Scale AI evaluation 37.5% labs.scale.com for Gemini 3 Pro on Humanity's Last Exam Full set — Scale AI evaluation (opens in a new tab) Archived copy (opens in a new tab) Humanity's Last Exam Accuracy Reported October 3, 2026 Benchmark leaderboard Scale AI Independent Record for Gemini 3 Pro on Humanity's Last Exam Full set — Scale AI evaluation Humanity's Last Exam Text only — Scale AI evaluation 37.7% labs.scale.com for Gemini 3 Pro on Humanity's Last Exam Text only — Scale AI evaluation (opens in a new tab) Archived copy (opens in a new tab) Humanity's Last Exam Accuracy Reported October 7, 2026 Benchmark leaderboard Scale AI Independent Record for Gemini 3 Pro on Humanity's Last Exam Text only — Scale AI evaluation MCP Atlas Public, April 2026 update 70.3% labs.scale.com for Gemini 3 Pro on MCP Atlas Public, April 2026 update (opens in a new tab) Pass rate Reported September 24, 2026 Benchmark leaderboard Scale AI Independent Record for Gemini 3 Pro on MCP Atlas Public, April 2026 update MMLU-Pro Original — Vals AI harness Configuration: Vals AI harness 90.1% vals.ai for Gemini 3 Pro on MMLU-Pro Original — Vals AI harness (opens in a new tab) Archived copy (opens in a new tab) Accuracy Reported September 1, 2026 Benchmark leaderboard Vals AI Independent Record for Gemini 3 Pro on MMLU-Pro Original — Vals AI harness MMMU-Pro No tools Effort: High Configuration: No tools 81.0% blog.google for Gemini 3 Pro (high) on MMMU-Pro No tools (opens in a new tab) MMMU-Pro accuracy Setting: Overall (Standard 10 options + Vision) Reported November 18, 2025 Launch post Google Self-reported Google DeepMind Record for Gemini 3 Pro on MMMU-Pro No tools MMMU-Pro Vals AI harness Effort: High Configuration: Vals AI harness 87.5% vals.ai for Gemini 3 Pro (high) on MMMU-Pro Vals AI harness (opens in a new tab) Archived copy (opens in a new tab) MMMU-Pro accuracy Setting: Standard (4 options) Reported September 1, 2026 Benchmark leaderboard Vals AI Independent Record for Gemini 3 Pro on MMMU-Pro Vals AI harness SWE-bench Multilingual — mini-swe-agent Configuration: mini-swe-agent harness 68.7% github.com for Gemini 3 Pro on SWE-bench Multilingual — mini-swe-agent (opens in a new tab) Archived copy (opens in a new tab) Resolved Tasks Reported February 13, 2026 Benchmark repository SWE-bench Independent SWE-bench Team Record for Gemini 3 Pro on SWE-bench Multilingual — mini-swe-agent SWE-bench Verified Effort: High 76.2% blog.google for Gemini 3 Pro (high) on SWE-bench Verified (opens in a new tab) Resolved Tasks Reported November 18, 2025 Launch post Google Self-reported Google DeepMind Record for Gemini 3 Pro on SWE-bench Verified SWE-bench Verified — Epoch AI harness Configuration: Epoch AI harness 72.9% epoch.ai for Gemini 3 Pro on SWE-bench Verified — Epoch AI harness (opens in a new tab) Archived copy (opens in a new tab) Resolved Tasks Reported October 7, 2026 Evaluated February 13, 2026 at 10:43 AM UTC Benchmark leaderboard Epoch AI Independent Record for Gemini 3 Pro on SWE-bench Verified — Epoch AI harness SWE-bench Verified — mini-swe-agent Effort: High Configuration: mini-swe-agent harness 69.6% github.com for Gemini 3 Pro (high) on SWE-bench Verified — mini-swe-agent (opens in a new tab) Archived copy (opens in a new tab) Resolved Tasks Reported February 26, 2026 Benchmark repository SWE-bench Independent SWE-bench Team Record for Gemini 3 Pro on SWE-bench Verified — mini-swe-agent SWE-bench Verified — mini-swe-agent Configuration: mini-swe-agent harness 74.2% github.com for Gemini 3 Pro on SWE-bench Verified — mini-swe-agent (opens in a new tab) Archived copy (opens in a new tab) Resolved Tasks Reported November 18, 2025 Benchmark repository SWE-bench Independent SWE-bench Team Record for Gemini 3 Pro on SWE-bench Verified — mini-swe-agent SWE-bench Verified — Vals AI mini-swe-agent Effort: High Configuration: mini-swe-agent harness 76.4% vals.ai for Gemini 3 Pro (high) on SWE-bench Verified — Vals AI mini-swe-agent (opens in a new tab) Archived copy (opens in a new tab) Resolved Tasks Reported September 1, 2026 Benchmark leaderboard Vals AI Independent Record for Gemini 3 Pro on SWE-bench Verified — Vals AI mini-swe-agent SWE-bench Pro Private — Scale AI SWE-Agent Configuration: SWE-Agent harness, 250 turns, uncapped cost 17.9% labs.scale.com for Gemini 3 Pro on SWE-bench Pro Private — Scale AI SWE-Agent (opens in a new tab) Archived copy (opens in a new tab) Public tasks resolved Reported October 7, 2026 Benchmark leaderboard Scale AI Independent Record for Gemini 3 Pro on SWE-bench Pro Private — Scale AI SWE-Agent SWE-bench Pro Public — Scale AI SWE-Agent Configuration: SWE-Agent harness, 250 turns, uncapped cost 43.3% labs.scale.com for Gemini 3 Pro on SWE-bench Pro Public — Scale AI SWE-Agent (opens in a new tab) Archived copy (opens in a new tab) Public tasks resolved Reported October 7, 2026 Benchmark leaderboard Scale AI Independent Record for Gemini 3 Pro on SWE-bench Pro Public — Scale AI SWE-Agent Terminal-Bench 2.0 Effort: High 54.2% blog.google for Gemini 3 Pro (high) on Terminal-Bench 2.0 (opens in a new tab) Terminal-Bench Accuracy Reported November 18, 2025 Launch post Google Self-reported Google DeepMind Record for Gemini 3 Pro on Terminal-Bench 2.0 Terminal-Bench 2.0 — Vals AI Terminus 2 Effort: High Configuration: Terminus-2 harness 55.1% vals.ai for Gemini 3 Pro (high) on Terminal-Bench 2.0 — Vals AI Terminus 2 (opens in a new tab) Archived copy (opens in a new tab) Terminal-Bench Accuracy Reported June 4, 2026 Benchmark leaderboard Vals AI Independent Record for Gemini 3 Pro on Terminal-Bench 2.0 — Vals AI Terminus 2 Toolathlon Original 36.4% toolathlon.xyz for Gemini 3 Pro on Toolathlon Original (opens in a new tab) Archived copy (opens in a new tab) Tasks completed Reported November 22, 2025 Benchmark leaderboard Toolathlon Independent Toolathlon Authors Record for Gemini 3 Pro on Toolathlon Original Vibe Code Bench 1.1 — OpenHands, October 2026 evaluation Effort: High Configuration: OpenHands harness 14.3% vals.ai for Gemini 3 Pro (high) on Vibe Code Bench 1.1 — OpenHands, October 2026 evaluation (opens in a new tab) Archived copy (opens in a new tab) Vibe Code Bench Score Reported October 6, 2026 Benchmark leaderboard Vals AI Independent Record for Gemini 3 Pro on Vibe Code Bench 1.1 — OpenHands, October 2026 evaluation τ³-bench τ³ — Airline Effort: High 80.5% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Gemini 3 Pro (high) on τ³-bench τ³ — Airline (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported March 2, 2026 Evaluated March 2, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Gemini 3 Pro on τ³-bench τ³ — Airline τ³-bench τ³ — Banking, Terminal retrieval, v1.0.1 grading Effort: High Configuration: Terminal retrieval 18.0% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Gemini 3 Pro (high) on τ³-bench τ³ — Banking, Terminal retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported March 2, 2026 Evaluated March 2, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Gemini 3 Pro on τ³-bench τ³ — Banking, Terminal retrieval, v1.0.1 grading τ³-bench τ³ — Retail Effort: High 75.9% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Gemini 3 Pro (high) on τ³-bench τ³ — Retail (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported March 2, 2026 Evaluated March 2, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Gemini 3 Pro on τ³-bench τ³ — Retail τ³-bench τ³ — Telecom Effort: High 91.0% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Gemini 3 Pro (high) on τ³-bench τ³ — Telecom (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported March 2, 2026 Evaluated March 2, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Gemini 3 Pro on τ³-bench τ³ — Telecom