Unverified50% confidenceReleaseExact time
The Nanbeige LLM Lab released Nanbeige4.3-33B, a 33B dense model.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/25/2026
First Seen
Valid until: 1/23/2027
Sources
Related Entities
Related Claims
UnverifiedThe Nanbeige4.3-33B model has been released on Hugging Face with 33 billion parameters.67% similarUnverifiedAccording to the Reddit community citing the Hugging Face model card, Nanbeige4.3-33B outperforms Qwen3.8-27B on a series of benchmarks.66% similarVerifiedQwen3.8-27B delivers impressive performance in specific scenarios such as front-end display, and may be one of the best-performing locally deployed large models currently available, showing a significant capability improvement over the Meta 30B model and Qwen3.6-27B.66% similarUnverifiedThe author noted that other models previously tested, such as 35B and 3B variants, actually performed worse than the Q2 version of Qwen3 27B on the Kanban task.61% similarUnverifiedXingchen 4.0 29B is a domestically developed open-source model that adopts a MoE architecture with 29B total parameters, activating only 4B parameters at a time.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1097831API
curl https://kongchang.com/api/v1/knowledge/claims/1097831MCP
get_claim(id=1097831)