Leading open models graded on the OMSF Openness Ladder, plus the full Hugging Face catalog with computed VRAM estimates, license and access analysis.
The Leaderboard is built for one decision: given your hardware and your needs, which open-weight model should you actually deploy? Each entry carries an OMSF openness grade, an estimated VRAM footprint for common quantization levels, and a plain-language read on its license and access restrictions.
The model catalog syncs daily from Hugging Face, so the list stays current without manual updates. If a model you need is missing a grade, the Reports Library holds full OMSF audit reports and the blog explains the reasoning behind the grades.