Gemma 3 12B benchmark results
5 results on 5 benchmarks from 4 sources.
Also listed as: gemma-3-12b-it · Gemma-3-12b-it (Prompt) · Gemma-3-12B-it
All benchmark results
| Benchmarks | Score | Rank | Setting | Source type | Date |
|---|---|---|---|---|---|
| LMArena Text | 1342 | 231 | Overall, style control on; 3829 votes; 95% CI 1332–1351“gemma-3-12b-it” | Official leaderboard | Oct 8, 2026 |
| GPQA Diamond | 39.5% | 240 | Epoch AI run, accuracy | Independent run | Aug 28, 2026 |
| MMLU-Pro | 60.6% | 135 | Overall accuracy; data source: Self-Reported; 12B“Gemma-3-12B-it” | Vendor-reported | Mar 11, 2026 |
| AIME (OTIS Mock AIME 2024–2025) | 16.7% | 209 | Epoch AI run, OTIS Mock AIME 2024–2025 | Independent run | Aug 28, 2026 |
| BFCL v4 (function calling) | 30.4% | 66 | Overall accuracy; prompt-based“Gemma-3-12b-it (Prompt)” | Official leaderboard | Oct 17, 2026 |
FAQ
Where do these Gemma 3 12B scores come from?
From Epoch AI Benchmarking Hub, Berkeley Function Calling Leaderboard (BFCL), LMArena leaderboard dataset, and MMLU-Pro leaderboard (TIGER-Lab). Each row links to the original leaderboard, dataset or report, and vendor-reported numbers are labelled.