Bonsai 27B Runs Locally on Mobile Phones
dair_ai · x · 2026-07-15
PrismML has teased/released Bonsai 27B, claiming it to be the first 27B-scale multimodal model capable of running on mobile phones.
The core highlight is compressing a traditionally unfeasible model size for local deployment down to mobile devices and laptops. It emphasizes support for multi-step reasoning, structured tool calling, long-context workflows, and more consistent agentic loops. The quantized sizes are detailed as follows:
- Ternary Bonsai 27B: 5.9 GB, 1.71 effective bits/weight, targeting laptop-level quality
- 1-bit Bonsai 27B: 3 GB
Regarding performance, reports indicate it hits up to 163 tok/s (1-bit) / 134 tok/s (ternary) on an NVIDIA GeForce RTX 5090, and 87 tok/s (1-bit) / 58 tok/s (ternary) on an M5 Max.
Related event: Bonsai 27B: The first 27B model that runs on phones(15 posts)→
More from Infra
- The shadow GPU capacity market is formalizing, with Meta selling excess compute to outside buyers — DavidLinthicum · 2026-09-11
- Engram's random reads don't suit SSDs; CPU-memory over NVLink could serve all 72 GPUs — bookwormengr · 2026-09-11
- 80% of the DIY LLM inference hype posters have already quit — it's brutally hard systems work — abhijithneil · 2026-09-11
- Hugging Face's Ultra Scale Playbook: a free book on training LLMs on GPU clusters — mdancho84 · 2026-09-11
- Is inference latency becoming the biggest bottleneck for production AI agents? — Euphoric_Sea632 · 2026-09-11
- LLM Serving Metrics Thread: Why TPOT and Uptime Make or Break User Experience — abhijithneil · 2026-09-11