Prism ML
PulseAugur coverage of Prism ML — every cluster mentioning Prism ML across labs, papers, and developer communities, ranked by signal.
- 2026-07-04 product_launch Prism ML has released new versions of its Bonsai 27B model, including 1-bit and ternary weight variants optimized for different hardware. source
1 day(s) with sentiment data
-
How to serve a 6GB 27B AI model: Configuration and optimization
Prism ML has detailed how to serve a 27 billion parameter model, Ternary Bonsai 2, which is compressed to just 6 GB. The process involves specific configurations for runtime environments and model parameters to achieve …
-
Bonsai 27B model compresses Qwen 3.6 to phone-size, excelling at code tasks
A new model called Bonsai 27B, developed by Prism ML, a spinout from Caltech, has been created by aggressively compressing the Qwen 3.6 model. This compression allows the model to fit on a phone, reducing its size from …
-
Prism ML releases ternary Bonsai models for fine-tuning
Prism ML has released its ternary Bonsai models, making them available for fine-tuning. The release includes example code and suggests using a high learning rate for the process. This development offers new options for …
-
Prism ML releases compact Bonsai Image 4B diffusion model
Prism ML has released Bonsai Image 4B, a text-to-image diffusion model that utilizes ternary weights for significant size reduction. The model is available in two versions: one optimized for Apple Silicon using MLX and …