Nanbeige releases Nanbeige4.3-33B agentic teacher model
Original title: Nanbeige/Nanbeige4.3-33B — new model trending #28 on Hugging Face
The lab says the 33B model leads most ~30B coding and cowork benchmarks and will distill a 3B model.
Nanbeige published Nanbeige4.3-33B on Hugging Face as the flagship teacher for the Nanbeige4.3 series, aimed at agentic coding and professional cowork tasks. The model card compares it with Qwen3.8-27B, Muse-Glimmer-30B, and Gemma4-31B at highest thinking effort under shared inference settings. Nanbeige reports leads on most listed benchmarks, including Terminal-Bench 2.1 at 80.9, while trailing on SWE-Bench Pro and GDPval. A Nanbeige4.3-3B distillation is described as upcoming.
Unconfirmed: Scores and the state-of-the-art claim are only from Nanbeige's model card.
- Terminal-Bench 4.0 is listed at 23.8 versus 6.3 for Qwen3.8-27B.
- SWE-Bench Pro is 60.3, behind Qwen3.8-27B at 61.7.
- AIME 2026 is 95.2; GPQA Diamond is 89.5, behind Gemma4-31B at 91.3.
- JobBench is said to beat the baseline in most of 35 occupations.
- Thinking is on by default; preserve_thinking is recommended off for general chat.
- Vendors
- Nanbeige
- AI models
- Nanbeige4.3-33B
From the original article
3 min read at huggingface.cotext-generation · 0 downloads · 172 likes · license apache-2. 0
By Nanbeige · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.