Scores updated · 11 tests both models report · How we compare
Every test, side by side
All 11 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Other results11 tests
- HarmBenchClaude 3 Sonnet by 61.495.834.4+61.4
- NarrativeQAMistral Instruct V0.3 (7B) by 60.511.171.6+60.5
- AIR-Bench 2024Claude 3 Sonnet by 52.184.732.5+52.1
- GSM8KClaude 3 Sonnet by 38.592.353.8+38.5
- MMLUClaude 3 Sonnet by 27.378.351+27.3
- bbqClaude 3 Sonnet by 23.29066.8+23.2
- naturalquestions_closedbookMistral Instruct V0.3 (7B) by 22.52.825.3+22.5
- simple_safety_testsClaude 3 Sonnet by 1910081+19
- OpenBookQAClaude 3 Sonnet by 12.891.879+12.8
- anthropic_red_teamClaude 3 Sonnet by 9.799.890.1+9.7
- XSTestMistral Instruct V0.3 (7B) by 6.885.892.6+6.8
Questions people ask
How do you compare the two?
We use the 11 benchmark tests both models have published scores on. None of them is in the eight capability areas we count, so this page lists them without a verdict. Each score is the one shown on the model's own page.