The Hugging Face Official Benchmarks map is updated daily, providing a snapshot of leaderboard standings across various AI models. As of October 20, 2026, there are 49 official benchmarks with 1,107 leaderboard entries. A recent change noted is SoMarkAI/OXR-1.0 taking the top spot on the olmOCR-bench, surpassing infly/Infinity-Parser2-Pro. The updates highlight that most entries are self-reported, and evaluation settings can vary. AI
IMPACT Provides ongoing tracking of model performance on established benchmarks, informing developers about relative strengths.
RANK_REASON This cluster reports on daily updates to an existing benchmark leaderboard, not a new model release or significant industry event.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →