MobileLLM-R1-140M: benchmark results
Meta MobileLLM-R1 140M is the final reasoning SFT checkpoint for mathematics, programming and scientific problems, separate from its Base release. The publisher specifies 15 layers, 9 attention heads, 3 key/value heads and shared input/output embeddings. Provider: Meta. Released 2025-09-12. Access: Open.
Aggregate Rating 1175 ± 38, rank #9739 of 10916 rated models, from 143 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Norwegian Knowledge - Idioms NO | 0.62 | MCC (x100) | 31.4 |
| EuroEval Lithuanian Common Sense Reasoning | 0.82 | Common Sense Reasoning Average Score (%) | 29.6 |
| EuroEval Estonian NLU - Grammar ET | 0.95 | Matthews correlation coefficient (%) | 25.7 |
| EuroEval Swedish NLU - ScaLA SV | 2.67 | Matthews correlation coefficient (%) | 22.7 |
| EuroEval Faroese NLU - ScaLA FO | 0.33 | Matthews correlation coefficient (%) | 22.6 |
| EuroEval Polish Common Sense Reasoning | 0.46 | Common Sense Reasoning Average Score (%) | 20.5 |
| EuroEval Icelandic Common Sense Reasoning | 0.45 | Common Sense Reasoning Average Score (%) | 18.6 |
| EuroEval Icelandic NLU - ScaLA IS | 0 | Matthews correlation coefficient (%) | 17.4 |
| EuroEval German NLU - ScaLA DE | 1.25 | Matthews correlation coefficient (%) | 15.1 |
| EuroEval Estonian Knowledge | 0.38 | Knowledge Average Score (%) | 15 |
| EuroEval Latvian Common Sense Reasoning | 0 | Common Sense Reasoning Average Score (%) | 14.9 |
| EuroEval Norwegian Common Sense Reasoning | 0.7 | Common Sense Reasoning Average Score (%) | 14.5 |
Interactive version: theaggregate.ai/model?slug=mobilellm-r1-140m · How It Works · Data refreshed daily, snapshot 2026-11-08.