Sao10K

Comparing 3 models from Sao10K.

Model Total ▼ Released Context SizeCreative writingRule followingUtilityMathematicsToolingLanguageLogic
Sao10K L3.1 70B Hanami x160.77%Sep 9, 2416k70B27.79%53.98%68.67%100.00%53.08%72.63%73.75%
Llama 3.1 Euryale 70B v2.255.08%Aug 26, 248k70B22.95%48.04%62.89%75.00%49.36%63.68%71.25%
Llama 3 Euryale 70B v2.151.22%Jun 14, 2416k70B17.87%40.19%59.17%80.00%44.62%60.17%79.38%
Model Performance
Cost vs Performance

Compares total benchmark cost against overall score for Sao10K models. Quadrant lines are drawn at the median values.