Apertus 70B
Apertus 70B ranks first on BLEnD.
33
Benchmarks with a score
1
Ranked first, out of five or more models
1
In the top three
Swiss AI Initiative Sep 2025 Apertus 70B
Open weights Flagship
No Epoch Capabilities Index score yet
- LEXam 34.7% #25 /31
- MLogiQA 40.5% #8 /8
- GPQA Main 30.6% #7 /7
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| BLEnD | 66.3% | #1 of 5 | Model card | |
| Multi-IF | 74.7% | #11 of 28 | Model card | |
| ARC-Challenge | 85% | #19 of 45 | Model card | |
| TruthfulQA | 61.2% | #13 of 28 | Model card | |
| PIQA | 81.9% | #18 of 33 | Model card | |
| WinoGrande | 73.3% | #29 of 42 | Model card | |
| MMLU | 69.6% | #97 of 127 | Model card | |
| HellaSwag | 78.1% | #36 of 47 | Model card | |
| BIG-Bench Hard | 64.2% | #55 of 71 | Model card | |
| MGSM | 64.3% | #42 of 52 | Model card | |
| IFEval | 75.2% | #108 of 126 | Model card | |
| AGIEval | 40.5% | #30 of 35 | Model card | |
| INCLUDE | 58.2% | #22 of 25 | Model card | |
| Global MMLU | 62.7% | #10 of 11 | Model card | |
| DROP | 50.8% | #39 of 40 | Model card | |
| SEA-HELM | 32.6 | #61 of 62 | Official leaderboard | |
| MLogiQA | 40.5% | #8 of 8 | Model card | |
| XCOPA | 69.8% | #2 of 2 | Model card | |
| ACPBench | 62.9% | #1 of 1 | Model card | |
| CulturalBench | 74.2% | #1 of 1 | Model card | |
| SwitzerlandQA | 67.2% | #1 of 1 | Model card | |
| XNLI | 45.3% | #1 of 1 | Model card |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| GSM8K | 77.6% | #63 of 87 | Model card | |
| MATH | 30.8% | #80 of 93 | Model card | |
| GPQA Main | 30.6% | #7 of 7 | Model card | |
| GSM8K-Platinum | 74.6% | #1 of 1 | Model card | |
| MathQA | 33.9% | #1 of 1 | Model card |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LEXam | 34.7% | #25 of 31 | Official leaderboard |
Safety and security
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| RealToxicityPrompts | 0.20% | #1 of 2 | Model card | |
| BBQ (Bias Benchmark for QA) | 67.4% | #22 of 24 | Model card | |
| HarmBench | 31.9% | #4 of 4 | Model card |
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.
Its system card
Every result in the longer documents the lab published with it. 29 of them aren't in the launch post.
Capabilities
- MMLU 69.6
- Global MMLU 62.7
- TruthfulQA 61.2
- HellaSwag 78.1
- HumanEval 73
- MBPP 47
- GSM8K 77.6
- MGSM 64.3
- MATH 30.8
- MathQA 33.9
- BIG-Bench Hard 64.2
- DROP 50.8
- ACPBench 62.9
- IFEval 75.2
- Multi-IF 74.7
- INCLUDE 58.2
- BLEnD 66.3
- CulturalBench 74.2
- SwitzerlandQA 67.2
- AGIEval 40.5
- ARC-Challenge 85
- GPQA Main 30.6
- GSM8K-Platinum 74.6
- MLogiQA 40.5
- RULER
Safety
- HarmBench 31.9
- RealToxicityPrompts 0.2
- LinguaSafe
Bias
- BBQ (Bias Benchmark for QA) 67.4
- ToxiGen 70.3