SmolLM2-1.7B — benchmark results

SmolLM2 1.7B base/pretrained checkpoint. Provider: HuggingFace. Released 2024-10-30. Access: Open.

Unified ELO 1245 ± 23, rank #1628 of 1776 rated models, from 68 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Fibble2 Arena0Win Rate (%)41.1
Fibble4 Arena0Win Rate (%)40.5
Fibble3 Arena0Win Rate (%)39.3
Fibble5 Arena0Win Rate (%)36.7
EuroEval Portuguese NLU - SST-2 PT70.07Sentiment classification Score (%)36.1
Open LLM Leaderboard - GPQA3.91Score34.3
EuroEval Spanish NLU - ScaLA ES1.83Linguistic acceptability Score (%)31.4
EuroEval Dutch NLU - DBRD74.28Sentiment classification Score (%)30.1
EuroEval Spanish NLU - MLQA ES46.34Reading comprehension Score (%)30.1
EuroEval Portuguese NLU - MultiWikiQA PT52.83Reading comprehension Score (%)30
EuroEval Italian NLU - SQuAD IT45.46Reading comprehension Score (%)29.3
EuroEval Icelandic Knowledge2.61Knowledge Average Score (%)28.9

Interactive version: theaggregate.ai/model?slug=smollm2-1-7b · How the rankings work · Data refreshed daily, snapshot 2026-07-22.