Nanbeige4.3-33B on BenchGecko · open weights Apache 2.0 dense 33B with 524,288 context
The Nanbeige LLM Lab released Nanbeige4.3-33B on 2026-10-24 under Apache 2.0. BenchGecko lists the dense 33B model with a 524,288 token context and vendor-reported scores.
What launched
The Nanbeige LLM Lab released Nanbeige4.3-33B on Hugging Face on 2026-10-24. It is a dense 33B parameter model with open weights under Apache 2.0, published as Nanbeige/Nanbeige4.3-33B. A base checkpoint, Nanbeige4.3-33B-Base, shipped alongside it, and the lab says a distilled 3B model is planned.
BenchGecko added the model to the Nanbeige provider page today.
Specs
From the Hugging Face model card and config:
- Parameters · 33B, dense
- Architecture · Llama style decoder, 52 layers
- Context · 524,288 tokens
- Inputs and output · text
- Modes · thinking mode and tool calling
- Weights · open, Apache 2.0
Pricing on BenchGecko
No hosted API price was found and the model is not on OpenRouter. BenchGecko leaves the USD price fields empty rather than inventing a list price.
Vendor-reported benchmarks
The lab reports these scores on the model card. BenchGecko records them as vendor claims. No independent scores or Gecko Tests are attached yet.
- GPQA Diamond · 89.5
- SWE-Bench Pro · 60.3
- AIME 2026 · 95.2
- AA-LCR · 84.0
- Terminal-Bench 2.1 · 80.9
- SciCode · 52.6
- GDPval · 84.8
- Tau3-Bench · 48.8
- SkillsBench · 61.1
- Toolathlon-Verified · 70.4
Sources
Source: BenchGecko
Written automatically from BenchGecko's tracked data for this period; every number comes from the changelog, model, mindshare and pricing datasets. Checked against the source data before publishing.