Latest Open Source Models (April 2026) — Optimized for 64GB RAM
Tests knowledge across 57 subjects including math, history, law, medicine, and more. Higher is better. 80%+ = excellent general knowledge, 70-80% = good, below 70% = average.
Measures code generation ability using 164 programming problems. Higher is better. 80%+ = excellent coder, 70-80% = good programming skills, below 70% = basic coding ability.
Select models from the table view to compare them here. Best values highlighted in green.