BOSS Zhipin's Nanbeige releases open-weights 33B model beating Qwen 27B on agentic benchmarks
Dan_Jeffries1 · x · 2026-10-26
Nanbeige, the research team of BOSS Zhipin, released Nanbeige4.3-33B with open weights, targeting SOTA agentic performance at 30B scale.
- Their published comparison shows it beating Qwen3.8-27B and Muse Glimmer-30B on most agentic coding and cowork benchmarks: Terminal-Bench 2.1 at 80.9 (vs 73.0), DeepSWE 65.5, Automation Bench 54.2, and YC-bench value of $1,025,437; it trails Qwen only on SWE-Bench Pro (60.3 vs 61.7) and GDPval.
- The strategy: build a strong teacher first, then distill smaller models starting with Nanbeige4.3-3B.
- The team says it plugs into mainstream agent harnesses for cost-efficient real-world tasks.
Thom Wolf notes China's open-weights roster now spans labs, cloud providers, quant funds, and even job boards.
More from Models
- Running a 125B model on 8GB VRAM with Strata: 900 t/s prompt processing, 30 t/s generation — BleynSpecnaz · 2026-10-26
- More OG Hugging Face profiles to follow: local AI, diffusion, kernels and transformer.js — ariG23498 · 2026-10-26
- 9 Hugging Face profiles worth following, from TRL agents to vLLM kernels — ariG23498 · 2026-10-26
- Terence Tao says OpenAI went 'code red' to beat Anthropic on Navier-Stokes, pausing $200 Pro signups — Hesamation · 2026-10-26
- LoopUno: 52-Minute LoRA Training Makes 3B LLM Draft 2 Tokens per Step Losslessly — LucaAmb · 2026-10-26
- OpenAI enterprise price cuts spark debate as users cite quota frustration and churn — teortaxesTex · 2026-10-26