Clover 2: a 27.8B base model with 7 LoRA experts routed per token
Kitani AI released Clover 2, a dense 27.8B base model kept separate from 7 trained LoRA expert adapters, with a trained router picking experts per token and falling back to the plain base when none fits. All components are public on Hugging Face.
Why it matters
Instead of merging experts into one static model, Clover 2 keeps them separate and routes per token, a small-model trick that could make one base model behave like seven specialists.
What you could build with it
A small team fine-tuning for several niches could train one LoRA adapter per niche on top of the shared 27.8B base, then let the router pick the right expert per token, getting specialist behavior without serving seven separate models.
Does it hold up?
Too early to judge: weights, router and a vLLM plugin are public, but the benchmark gains are the team's own reports and no independent runs exist yet.
Learn more
First spotted on hf: source.
More AI models
VoiceStudio: open-source, fully-local ElevenLabs alternative for voice cloning, dubbing, and audiobooksdebpalash/VoiceStudio is a desktop app that runs voice cloning, voice design, video dubbing, dictation, transcription…models · JEV 0.75Edge0-145B: open 145B MoE that runs on a single consumer GPUEdge0 open-sourced 145B and 120B latent-MoE models derived from Qwen3.8-Flash-Next, plus an inference framework that…models · JEV 0.74Strata: run a 125B model on a gaming PCFree, MIT-licensed local inference engine that runs Qwen3.8-Flash-Next (125B MoE, about 6B active per token) on…models · JEV 0.73Google AI Edge Gallery: offline speech recognition and synthesis on-deviceGoogle added fully offline speech recognition and text-to-speech to AI Edge Gallery, with ML Kit APIs on Android and…models · JEV 0.73Soup: fine-tune an 8B LLM on a 4 GB laptop GPU from one YAML fileAn open-source CLI that fine-tunes LLMs from a single YAML file, with layer streaming that trains an 8B model on a 4…models · JEV 0.73Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.