Moonshot AI publishes Moonlight-16B-A3B base model on Hugging Face
Moonshot AI published the Moonlight-16B-A3B base model on Hugging Face under the moonshotai organisation. Tags indicate it uses the DeepSeek V3 architecture, is MIT licensed, and has an associated paper (arxiv:2502.16982). This is the base (non-instruct) variant alongside the separately released instruct version. The source is a repository listing only; no capability claims or benchmark results are asserted.
PUBLISHED2025-02-22
OBSERVED2026-08-11
AGE1y
SOURCES1
- Repository: moonshotai/Moonlight-16B-A3B on Hugging Face
- Pipeline: text-generation, library: transformers
- Tags: deepseek_v3, safetensors, text-generation, conversational, custom_code
- Associated paper: arxiv:2502.16982
- License: MIT
- Source is a repository listing only; no benchmark results or capability claims are present
COMMUNITY
No curated reactions recorded for this event. Facts and takes are kept in separate layers — community context is added by hand, never blended into the record above.