Scores updated · 27 tests both models report · How we compare
Every test, side by side
All 27 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Other results27 tests
- MBPP+ERNIE 5.5 Flash Base by 14.367.281.5+14.3
- HumanEval+ERNIE 5.5 Flash Base by 12.869.582.3+12.8
- FullStackBenchERNIE 5.5 Flash Base by 11.343.354.6+11.3
- RepoQA (Pass@1)ERNIE 5.5 Flash Base by 10.78191.7+10.7
- WinoGrande (EM)ERNIE 5.5 Flash Base by 8.379.587.8+8.3
- LiveCodeBenchERNIE 5.5 Flash Base by 8.231.940.1+8.2
- BBH (EM, direct)ERNIE 5.5 Flash Base by 8.16977.1+8.1
- PreciseWikiQA (LLM-Judge)ERNIE 5.5 Flash Base by 7.35259.3+7.3
- HotpotQA (LLM-Judge)ERNIE 5.5 Flash Base by 5.551.857.3+5.5
- SimpleQA (LLM-Judge)ERNIE 5.5 Flash Base by 529.134.1+5
- GPQA (EM)ERNIE 5.5 Flash Base by 4.355.960.2+4.3
- LongBench-V2 (EM)ERNIE 5.5 Flash Base by 444.748.7+4
- HELMET-Rerank (NDCG@10)DeepSeek-V4-Flash-Base by 3.86056.2+3.8
- PopQA (LLM-Judge)ERNIE 5.5 Flash Base by 3.548.752.2+3.5
- MathBench-CN (LLM-Judge)ERNIE 5.5 Flash Base by 2.87880.8+2.8
- HELMET-RAG (SubEM)ERNIE 5.5 Flash Base by 2.668.971.5+2.6
- MMMLU-lite (EM)ERNIE 5.5 Flash Base by 2.578.581+2.5
- ChineseSimpleQA (LLM-Judge)DeepSeek-V4-Flash-Base by 2.375.373+2.3
- HELMET-ICL (EM)ERNIE 5.5 Flash Base by 2.183.885.9+2.1
- BBH (EM, CoT) (Pass@1 with chain-of-thought)ERNIE 5.5 Flash Base by 1.787.589.2+1.7
- InfiniteBench (Avg)DeepSeek-V4-Flash-Base by 1.54846.5+1.5
- MGSM (EM)ERNIE 5.5 Flash Base by 1.385.787+1.3
- RepoBench (EM)ERNIE 5.5 Flash Base by 1.249.150.3+1.2
- PIQA (EM)ERNIE 5.5 Flash Base by 19596+1
- MultiLoKo (LLM-Judge)tie42.643.3tie
- CMMLU (EM)tie90.490tie
- INCLUDE-lite (EM)tie79.879.5tie
Questions people ask
How do you compare the two?
We use the 27 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.