The two are evenly matched.
Scores updated · 22 tests both models report · How we compare
Where each one wins
Tests won in each of the one area where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
Reasoning rests on a single test.
Every test, side by side
All 22 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Other results21 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- HumanEvalGLM-4.5-Base by 34.943.378.2+34.9
- Chinese SimpleQA (C-SimpleQA)GLM-4.5-Base by 2048.568.5+20
- CRUXEval-O (output prediction)GLM-4.5-Base by 1849.867.8+18
- MGSMGLM-4.5-Base by 17.763.681.3+17.7
- MATHGLM-4.5-Base by 17.643.461+17.6
- MBPPGLM-4.5-Base by 16.66581.6+16.6
- CRUXEval-I (input prediction)GLM-4.5-Base by 1652.568.5+16
- MMMLUGLM-4.5-Base by 15.36479.3+15.3
- MMLU-ProGLM-4.5-Base by 12.351.463.7+12.3
- MMLUGLM-4.5-Base by 12.175.687.7+12.1
- GSM8KGLM-4.5-Base by 10.979.290.1+10.9
- CMathGLM-4.5-Base by 10.678.789.3+10.6
- MMLU-ReduxGLM-4.5-Base by 8.777.986.6+8.7
- BBHGLM-4.5-Base by 7.478.886.2+7.4
- C-EvalGLM-4.5-Base by 4.481.485.8+4.4
- ARC-ChallengeGLM-4.5-Base by 4.192.296.3+4.1
- HellaSwagGLM-4.5-Base by 3.187.190.2+3.1
- DROPGLM-4.5-Base by 2.880.182.9+2.8
- CMMLUGLM-4.5-Base by 2.58486.5+2.5
- WinoGrandeDeepSeek-V2 by 1.186.385.2+1.1
- PIQAtie83.984.7tie
Questions people ask
Which is better, DeepSeek-V2 or GLM-4.5-Base?
Neither wins more tests than the other in any of the one area where both have results.
How do you compare the two?
We use the 22 benchmark tests both models have published scores on. The verdict counts the 1 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 21 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.