Scores updated · 20 tests both models report · How we compare
Every test, side by side
All 20 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Other results20 tests
- HumanEvalStep3.5 Flash Base by 14.181.167+14.1
- MultiPL-E MBPPStep3.5 Flash Base by 10.15847.9+10.1
- MBPP+ (EvalPlus-augmented)Step3.5 Flash Base by 8.270.662.4+8.2
- Chinese SimpleQA (C-SimpleQA)Step3.5 Flash Base by 5.163.258.1+5.1
- MultiPL-E HumanEvalStep3.5 Flash Base by 4.767.763+4.7
- WinoGrandeStep3.5 Flash Base by 3.379.175.8+3.3
- SimpleQAStep3.5 Flash Base by 3.231.628.4+3.2
- HumanEval+Step3.5 Flash Base Midtrain by 37275+3
- C-EvalStep3.5 Flash Base by 2.489.687.2+2.4
- MMLUStep3.5 Flash Base by 2.485.883.4+2.4
- GPQA (unspecified)Step3.5 Flash Base Midtrain by 2.141.743.8+2.1
- CMMLUStep3.5 Flash Base by 288.986.9+2
- MMLU-ReduxStep3.5 Flash Base by 289.287.2+2
- MMLU-ProStep3.5 Flash Base Midtrain by 1.362.363.6+1.3
- MATHStep3.5 Flash Base by 1.166.865.7+1.1
- BBHtie88.287.3tie
- HellaSwagtie90.291tie
- GSM8Ktie88.288.9tie
- MBPPtie79.479tie
- SuperGPQAtie4141.2tie
Questions people ask
How do you compare the two?
We use the 20 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.