Kapitsa Clover
Clover 35B-A3B (4-bit, 19.5 GB) running locally on Apple Silicon through Kapitsa. DiffuSpace's block-diffusion MoE tuned from Qwen3.6-35B-A3B.
kapitsa serve --model clover
Speed
| Mac | Prefill (tok/s) | Decode (tok/s) |
|---|---|---|
| M5 Max 128 GB | 5,981 | 304 |
Architecture
Block-diffusion MoE, 35B total parameters, 3B active per token, 4-bit quantization.
Weights
Model tree for nace-ai/kapitsa-clover
Base model
Qwen/Qwen3.6-35B-A3B Finetuned
Diffuspace/Clover-35B-A3B-Preview