Up to four models
Gemma 3 12B vs Llama 3.1-70B
On the index (GPT-4 = 100), Gemma 3 12B scores 107 and Llama 3.1-70B 106. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
10789–124
WeightsOpen
ReleasedMar 2025
Price in / out$0.050 / $0.15
Context131K
Core tests taken1 of 9
10689–122
WeightsOpen
ReleasedJul 2024
Price in / out$0.72 / $0.72
Context131K
Core tests taken1 of 9
Gemma 3 12B and Llama 3.1-70B are too close to call (1 point apart, within the uncertainty), ahead on 0 of 1 shared core tests, and costs 9.6× less.
Test by testthird-party results
Every result as a table
| Test | Gemma 3 12B | Llama 3.1-70B |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 39.5% | 44.2% |
| Maths | ||
| AIME-style maths (OTIS mock) | 16.7% | 3.6% |
| Agents | ||
| DTBench | 48.8% | 60.0% |
| LMCA | 4.5% | 14.8% |
Tests at least two of these models have taken. Bold rows are the core tests.