SmolLM2-360M
SmolLM2-360M.
15
Benchmarks with a score
0
Ranked first, out of five or more models
0
In the top three
SmolLM2-360M
Open weights Small
No Epoch Capabilities Index score yet
- Angry Tweets 4.99% #147 /149
- FoQA 4.12% #85 /91
- GermanQuAD 20.2% #79 /88
How does it do against its rivals?
SmolLM2-360M beats Gemma 2 27B on 1 of the 8 benchmarks they share. Qwen3-0.6B does best against it, ahead on 12 of 13.
-
Gemma 2 27B 1 ahead, 7 behind Compare
-
OLMo 2 32B Instruct 1 ahead, 8 behind Compare
-
Qwen3-8B 1 ahead, 8 behind Compare
-
Gemma 3 1B PT 1 ahead, 11 behind Compare
-
Llama 3.1-8B 0 ahead, 14 behind Compare
-
Qwen3-0.6B 0 ahead, 12 behind, 1 tied Compare
Sources: Official leaderboards, Epoch AI (CC BY 4.0) and labs’ model cards
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 | |
|---|---|---|---|---|
| EuroEval Faroese Grammatical Error Correction | 0.09% | #60 of 71 | Official leaderboard | |
| GermanQuAD | 20.2% | #79 of 88 | Official leaderboard | |
| OrangeSum | 25% | #31 of 34 | Official leaderboard | |
| FoQA | 4.12% | #85 of 91 | Official leaderboard | |
| Uncheatable Eval | 10.1 | #31 of 33 | Official leaderboard | |
| GuiaCat | 29.4% | #33 of 35 | Official leaderboard | |
| FoSent | 2.97% | #141 of 147 | Official leaderboard | |
| EuroEval ERRNews | 13.6 | #73 of 76 | Official leaderboard | |
| MMLU-Pro | 11.4% | #209 of 212 | Official leaderboard | |
| Angry Tweets | 4.99% | #147 of 149 | Official leaderboard | |
| Dutch Book Reviews Dataset | 14.6% | #104 of 105 | Official leaderboard | |
| Pedagogy Benchmark | 25.9% | #214 of 216 | Official leaderboard | |
| SEND Pedagogy Benchmark | 26.1% | #229 of 230 | Official leaderboard | |
| BLiMP-NL | 0.46% | #21 of 21 | Official leaderboard | |
| Dutch CoLA | 0.46% | #21 of 21 | Official leaderboard |
Who measured it?
Independent runs cover all 15 of its benchmarks.
- An independent run only 15
Sources: Official leaderboards, Epoch AI (CC BY 4.0) and labs’ model cards