Gemini 3 Pro benchmark results
24 results on 11 benchmarks from 6 sources.
Closed Released: Nov 18, 2025 Updated Oct 22, 2026
Also listed as: gemini-3-pro · gemini-3-pro-preview · Gemini 3 Pro Preview · live-SWE-agent + Gemini 3 Pro Preview (2025-11-18) · Gemini 3 Pro Preview (2025-11-18) · Gemini-3-Pro-Preview (Prompt) · Gemini-3-Pro-Preview (FC) · Gemini-3-Pro(11/25)
All benchmark results
| Benchmarks | Score | Rank | Setting | Source type | Date |
|---|---|---|---|---|---|
| LMArena Text | 1486 | 19 | Overall, style control on; 42078 votes; 95% CI 1482–1489“gemini-3-pro” | Official leaderboard | Oct 8, 2026 |
| LMArena WebDev | 1440 | 74 | Overall; 14090 votes; 95% CI 1431–1448“gemini-3-pro” | Official leaderboard | Oct 8, 2026 |
| LMArena Vision | 1289 | 13 | Overall, style control on; 13620 votes; 95% CI 1282–1297“gemini-3-pro” | Official leaderboard | Oct 2, 2026 |
| GPQA Diamond | 92.6% | 24 | Epoch AI run, accuracy | Independent run | Nov 19, 2025 |
| MMLU-Pro | 90.1% | 2 | Overall accuracy; data source: Self-Reported“Gemini-3-Pro(11/25)” | Vendor-reported | Mar 11, 2026 |
| SWE-bench Verified | 72.9% | 21 | Epoch AI run (Epoch scaffold) | Independent run | Feb 13, 2026 |
| SWE-bench Verified | 77.4% | 3 | agent: live-SWE-agent“gemini-3-pro-preview” | Official leaderboard | Nov 20, 2025 |
| SWE-bench Verified | 74.2% | 14 | agent: mini-SWE-agent“gemini-3-pro-preview” | Official leaderboard | Nov 18, 2025 |
| AIME (OTIS Mock AIME 2024–2025) | 91.4% | 65 | Epoch AI run, OTIS Mock AIME 2024–2025 | Independent run | Nov 19, 2025 |
| Humanity's Last Exam | 37.5% | 8 | Compiled by Epoch AI | Third-party compilation via Epoch AI | Nov 18, 2025 |
| τ²-bench | 80.5% | 6 | airline domain, pass^1; effort: high; submitted by Sierra; v1.0.1 | Official leaderboard | Mar 2, 2026 |
| τ²-bench | 18.0% | 21 | banking_knowledge domain, pass^1; effort: high; submitted by Sierra; v1.0.1 | Official leaderboard | Mar 2, 2026 |
| τ²-bench | 75.9% | 15 | retail domain, pass^1; effort: high; submitted by Sierra; v1.0.1 | Official leaderboard | Mar 2, 2026 |
| τ²-bench | 91.0% | 10 | telecom domain, pass^1; effort: high; submitted by Sierra; v1.0.1 | Official leaderboard | Mar 2, 2026 |
| Terminal-Bench 2.0 | 62.2% | 2 | Terminal-Bench 2.0; agent: CodeBrain-1 | Third-party compilation via Epoch AI | Feb 5, 2026 |
| Terminal-Bench 2.0 | 69.4% | 26 | Terminal-Bench 2.0; agent: Ante | Official leaderboard via Epoch AI | Nov 18, 2025 |
| Terminal-Bench 2.0 | 65.2% | 32 | Terminal-Bench 2.0; agent: SageAgent | Official leaderboard via Epoch AI | Nov 18, 2025 |
| Terminal-Bench 2.0 | 62.3% | 42 | Terminal-Bench 2.0; agent: CodeBrain-1.5 | Official leaderboard via Epoch AI | Nov 18, 2025 |
| Terminal-Bench 2.0 | 61.8% | 43 | Terminal-Bench 2.0; agent: II-Agent | Official leaderboard via Epoch AI | Nov 18, 2025 |
| Terminal-Bench 2.0 | 61.1% | 46 | Terminal-Bench 2.0; agent: Droid | Official leaderboard via Epoch AI | Nov 18, 2025 |
| Terminal-Bench 2.0 | 56.9% | 58 | Terminal-Bench 2.0; agent: Terminus 2 | Official leaderboard via Epoch AI | Nov 18, 2025 |
| Terminal-Bench 2.0 | 56.0% | 59 | Terminal-Bench 2.0; agent: Letta Code | Official leaderboard via Epoch AI | Nov 18, 2025 |
| BFCL v4 (function calling) | 72.5% | 3 | Overall accuracy; prompt-based“Gemini-3-Pro-Preview (Prompt)” | Official leaderboard | Oct 17, 2026 |
| BFCL v4 (function calling) | 68.1% | 7 | Overall accuracy; native function calling (FC)“Gemini-3-Pro-Preview (FC)” | Official leaderboard | Oct 17, 2026 |
FAQ
Where do these Gemini 3 Pro scores come from?
From Epoch AI Benchmarking Hub, Berkeley Function Calling Leaderboard (BFCL), LMArena leaderboard dataset, MMLU-Pro leaderboard (TIGER-Lab), SWE-bench leaderboards (swebench.com), and τ²-bench leaderboard submissions. Each row links to the original leaderboard, dataset or report, and vendor-reported numbers are labelled.