Scores updated · 17 tests both models report · How we compare
Every test, side by side
All 17 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Other results17 tests
- C-EvalStep3.5 Flash Base by 17.172.589.6+17.1
- CMMLUStep3.5 Flash Base by 15.273.788.9+15.2
- SimpleQAStep3.5 Flash Base by 14.517.131.6+14.5
- Chinese SimpleQA (C-SimpleQA)Step3.5 Flash Base by 12.850.463.2+12.8
- MBPPStep3.5 Flash Base by 1168.479.4+11
- MMLU-ProLlama 3.1 Instruct 405B by 1173.362.3+11
- GSM8KLlama 3.1 Instruct 405B by 8.696.888.2+8.6
- HumanEvalLlama 3.1 Instruct 405B by 7.98981.1+7.9
- MultiPL-E MBPPLlama 3.1 Instruct 405B by 7.765.758+7.7
- MultiPL-E HumanEvalLlama 3.1 Instruct 405B by 7.575.267.7+7.5
- MATHLlama 3.1 Instruct 405B by 773.866.8+7
- WinoGrandeLlama 3.1 Instruct 405B by 6.185.279.1+6.1
- BBHStep3.5 Flash Base by 5.382.988.2+5.3
- MMLU-ReduxStep3.5 Flash Base by 386.289.2+3
- MMLULlama 3.1 Instruct 405B by 2.888.685.8+2.8
- MBPP+ (EvalPlus-augmented)Llama 3.1 Instruct 405B by 2.47370.6+2.4
- HellaSwagStep3.5 Flash Base by 189.290.2+1
Questions people ask
How do you compare the two?
We use the 17 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.