Open-weight model intelligence · benchmark map
Open-Weight LLM × Domain Proficiency
Which downloadable-weights model to reach for, by professional domain. Every cell is a 0-5 proficiency score grounded in public benchmarks, with the underlying evidence and citations one click away. Click any cell.
Top pick by domain
Model × domain matrix · sorted by model
verified axis (adversarially fact-checked)
confidence: solid=high, faint=low
Rreasoning model