Nanbeige Releases Nanbeige4.3-33B, Claiming Edge Over Qwen and Gemma on Most Benchmarks
Nanbeige, the research team of BOSS Zhipin, released Nanbeige4.3-33B, an open-weight model it describes as state-of-the-art in the roughly 30B parameter class. The team claims it leads on most benchmarks, including agentic coding and cowork tasks, outperforming Qwen3.8-27B, Muse-Glimmer-30B and Gemma4-31B. Hugging Face co-founder Thomas Wolf said it surpasses Qwen 27B on most agent benchmarks. The weights are available under Apache 2.0 with a 512K-token context window, and a smaller Nanbeige4.3-3B is planned as a distillation target.
IllustrationEditorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page reads the event directly, while its address stays stable when the title changes.
- Summary covers the current reports
Cross-source coverage
Reporting timeline
BOSS Zhipin's Nanbeige Releases Open-Weight Nanbeige4.3-33B Agent Model
Nanbeige, the research team of BOSS Zhipin, released an open-weight model called Nanbeige4.3-33B. Thomas Wolf, co-founder and CSO of Hugging Face, said it surpasses Qwen 27B on most Agent benchmarks. The team claims it is the strongest Agent-performing open-source model in the roughly 30B class, with standout results on benchmarks such as Agent coding. Nanbeige also said the model will later serve as a teacher model to distill Nanbeige4.3-3B.
Read sourceNanbeige Releases Nanbeige4.3-33B to Push Small-Agent Performance Ceiling
Nanbeige announced the release of Nanbeige4.3-33B, a new model the company says is aimed at pushing the performance ceiling of small agents. In the announcement, Nanbeige stated that its strategy in Nanbeige4.3 begins with this goal, framing the release as an effort to improve what compact agent systems can achieve. The post did not provide further technical details, benchmark results, or availability information in the visible text. The announcement was shared via a repost by ClementDelangue.
Nanbeige Releases Nanbeige4.3-33B, Claiming SOTA in ~30B Open Model Class
Nanbeige released Nanbeige4.3-33B, which it describes as a new state-of-the-art open model in the roughly 30B parameter class. According to the announcement, the model leads on most benchmarks and shows strong performance in both agentic coding and cowork tasks. The post states that Nanbeige4.3-33B outperforms Qwen3.8-27B, Muse-Glimmer-30B, and Gemma4-31B on most benchmarks. Nanbeige also said a smaller model, Nanbeige4.3-3B, is planned for the future. The claims are attributed to the company's own announcement and were not independently verified in the source item.
Read sourceShow 2 older updatesHide older updates
Nanbeige Releases Nanbeige4.3-33B Open-Source Model
Nanbeige has released Nanbeige4.3-33B, which it describes as a new state-of-the-art open-source model in the roughly 30B parameter class. According to the source, the model leads on most benchmarks and shows strong performance in agentic coding and collaborative tasks. The report also says it outperforms Qwen3.8-27B, Muse-Glimmer-30B, and Gemma4-31B on most benchmarks. Nanbeige additionally plans to launch a smaller model, Nanbeige4.3-3B, in the future.
Read sourceNanbeige Releases Nanbeige4.3-33B, Claiming Edge Over Qwen 3.8 27b and Gemma
Nanbeige has released Nanbeige4.3-33B, a new model that the source says delivers surprisingly strong coding and agent performance for its size, beating similarly sized Qwen 3.8 27b and Gemma models on most benchmarks. The weights are openly available under the Apache 2.0 license, and the model supports a 512K-token context window. The post frames the release as a notably strong result for a model of this scale and calls it a really cool release. No independent verification of the benchmark comparisons is provided in the source item.