The Hugging Face Official Benchmarks map has been updated as of November 19, 2026, reflecting changes across 63 official benchmarks and 1,462 leaderboard entries. Several models have achieved new top rankings, including PromptforceAI/niti-gate-vegaml-4b on safeagents-judge-bench, thinkingmachines/Inkling-2-Small on SWE-bench_Pro and SWE-bench_Multimodal, and zai-org/GLM-5.3 on terminal-bench. The update also introduced three new official benchmarks: ParseBench-2, multiloko, and IFBench_test, along with 33 new leaderboard entries. AI
IMPACT Provides updated performance metrics for AI models, influencing future development and selection.
RANK_REASON The item reports on updates to an official benchmark leaderboard, including new top-ranking models and new benchmarks, which falls under research and evaluation of AI models. [lever_c_demoted from research: ic=1 ai=1.0]
- Hugging Face
- IFBench_test
- moonshotai/Kimi-K3.1-Pro
- multiloko
- ParseBench-2
- PCL-RSI-lab/Seshat-agent
- PromptforceAI/niti-gate-vegaml-4b
- thinkingmachines/Inkling-2-Small
- VIDRAFT
- vllm-sr/d3-ultra
- zai-org/GLM-5.3
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →