Back to Global AI & Tech NewsAI Models
PrismML hopes its tiny LLM will change how we all use AI
Source: TechCrunch AI
Executive Summary
Startup PrismML is developing remarkably small reasoning large language models capable of running locally on PCs and smartphones. The company released Bonsai 2 27B, compressing an open-source Alibaba model down to 5.9 GB while retaining 98% of its original benchmark performance.
PrismML is betting that capable, high-performing reasoning large language models don't have to be massive. The startup is engineering reasoning models small enough to fit on PCs and smartphones.
On Thursday, PrismML released Bonsai 2 27B, which compresses Alibaba's open-source Qwen3.8 27B down to 5.9 GB—a 9x to 10x reduction in memory usage compared to the original. Founded by Caltech researchers and led by compression expert Babak Hassibi, the startup reports that Bonsai 2 matches 98% of Qwen’s aggregate benchmark scores with virtually no performance loss.
Confirm and follow the full story at the original source:TechCrunch AI