Roadmap
What we're building.
Model compression work completed, in progress, and planned at Minutus.
Done
- Apertus-8B-Instruct (3-bit)
- Qwen3-4B (3-bit; 2-bit research preview)
- Qwen3-8B (3-bit)
In progress
- Qwen3.8-27B
- Optimized inference runtime for compressed models
Planned
- Gemma 4
- Muse Glimmer
- Mistral Large 3
- GLM-5.2
Need a model that is not listed?
Request a pilot