Observed Signal · Sep 17, 2026 · Product Launch · Source: techcrunch · Impact: 4/5 · Sentiment: Positive
PrismML launches tiny Bonsai 2 27B LLM for on-device AI
AI startup PrismML has released Bonsai 2 27B, a compressed large language model that fits on PCs and potentially high-end smartphones. The model compresses Alibaba's Qwen3.8 27B model to 5.9 GB, a 9x to 10x reduction in memory, while retaining 98% of the original's benchmark performance. Founded by Caltech researchers and led by CEO Babak Hassibi, PrismML has raised a $22.25 million seed round from Khosla Ventures, Cerberus Capital, and Caltech. The company's ternary weight compression technique simplifies model weights to just +1, -1, or 0. PrismML plans to apply this technology to larger models in the coming months. The startup is reportedly in talks with Apple, though this has not been confirmed.
Major technical release from an AI startup with potential to shift AI inference to on-device, impacting advertising and marketing tech by enabling private, low-cost AI at scale.
Track Apple Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- PrismML released Bonsai 2 27B, compressing Qwen3.8 27B to 5.9 GB, a 9-10x memory reduction.
- Bonsai 2 achieves 98% of Qwen's aggregate benchmark scores, up from 95% for the first Bonsai.
- The company raised a $22.25 million seed round from Khosla Ventures, Cerberus Capital, and Caltech.
- Original Bonsai models have been downloaded over 11 million times; smaller models another 2.6 million times.
- PrismML is in talks with Apple for potential integration, as reported by CNBC.
Connected Companies & Entities
4 Entities mapped“Rumored to be in talks with Apple for AI compression on iPhone....”
“Backed by investors Khosla Ventures, Cerberus Capital, and Caltech....”
“Bonsai 2 compresses Qwen3.8 27B, an open source model from Alibaba....”
“Ion Stoica, adviser, is co-founder of Databricks....”
Ontology Mapping & Concepts
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.
