Up to four models
Claude 2.1 vs Llama 2
On the index (GPT-4 = 100), Claude 2.1 scores 96 and Llama 2 95. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
Claude 2.1 and Llama 2 are too close to call (1 point apart, within the uncertainty), ahead on 1 of 1 shared core tests.
Test by testthird-party results
Claude 2.1
Llama 2
Every result as a table
| Test | Claude 2.1 | Llama 2 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 33.0% | 26.3% |
| MMLU (HELM runs) | 73.5% | 69.5% |
| Maths | ||
| AIME-style maths (OTIS mock) | 1.9% | 0.0% |
| Agents | ||
| DTBench | 50.9% | 41.6% |
Tests at least two of these models have taken. Bold rows are the core tests.