Nanbeige/Nanbeige4.3-33B
AI model · https://huggingface.co/Nanbeige/Nanbeige4.3-33B
First seen by Oriel on .
AI summary
Nanbeige4.3-33B is a 33B-parameter teacher model for the Nanbeige4.3 series, designed for coding and professional agent tasks, with reported leading results across many benchmarks in its parameter class.
- The model scores highest among the compared models on most listed agentic coding and cowork benchmarks, including Terminal-Bench 2.1, DeepSWE, and Automation Bench.
- Its tokenizer supports configurable reasoning and tool use; the authors recommend XML tool-call formatting and retaining prior reasoning for multi-turn tool-use workflows.
- The model supports up to 524,288 tokens of context and can be run with Transformers, SGLang, or vLLM.
- Nanbeige4.3-33B is intended both for immediate deployment and as a foundation for distilling agent capabilities into smaller models such as Nanbeige4.3-3B.
- Benchmark results use varying harnesses and evaluation setups, which are specified in the model documentation.
#large language models #llm agents #model distillation #inference
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Likes
Total likes reported by the source.
- Hugging Face Models: 1,057, latest observation ; +925 over observed period since
Downloads (30d)
Rolling 30-day downloads. A change in this value is not new downloads during the observed period.
- Hugging Face Models: 20,062, latest observation ; +20,062 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hugging Face Models | Nanbeige/Nanbeige4.3-33B | https://huggingface.co/Nanbeige/Nanbeige4.3-33B Read discussion |