gpt-sw3-6.7B-v2: benchmark results

Provider: AI Sweden. Released 2023-04-28. Access: Open.

Unified ELO 1251 ± 20, rank #2805 of 2928 rated models, from 48 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
EuroEval Danish NLU - Multi Wiki QA DA72.38Reading comprehension Score (%)57.2
EuroEval Swedish NLU - Multi Wiki QA SV69.48Reading comprehension Score (%)56.9
EuroEval Icelandic NLU - ScaLA IS0.69Linguistic acceptability Score (%)33
EuroEval Swedish NLU - ScaLA SV9.94Linguistic acceptability Score (%)30.8
EuroEval Icelandic Knowledge2.32Knowledge Average Score (%)27.8
EuroEval Faroese NLU - FoSent5.1Sentiment classification Score (%)25.1
EuroEval Danish Knowledge - Danish Citizen Tests15.59MCC (x100)22.1
EuroEval Icelandic NLU - Hotter and Colder Sentiment7.47Sentiment classification Score (%)21.9
EuroEval Danish Knowledge9.63Knowledge Average Score (%)21.8
EuroEval Danish NLU23.36NLU Average Score (%)21.6
EuroEval Norwegian Knowledge - NRK Quiz QA5.61MCC (x100)21.1
EuroEval Danish Knowledge - Danske Talemaader3.68MCC (x100)20.1

Interactive version: theaggregate.ai/model?slug=gpt-sw3-6-7b-v2 · How It Works · Data refreshed daily, snapshot 2026-09-23.