← Founder Notes
Archive

A 27 billion parameter model just moved onto a laptop. prism ml's bonsai 27b, out september 19,…

Yethikrishna ROriginal on Threads

a 27 billion parameter model just moved onto a laptop. prism ml's bonsai 27b, out september 19, packs a full reasoning model into 1-bit weights, cutting the download from 54 gigabytes to something a flagship phone can run.

the model shrank, not the reasoning.

Context

PrismML's announcement dated July 14, 2026 introduces Bonsai 27B, based on Qwen3.6 27B, in two variants. Ternary Bonsai 27B uses ternary weights at 1.71 effective bits per weight, is 5.9 GB, and runs on an everyday laptop. 1-bit Bonsai 27B uses binary weights at 1.125 effective bits per weight, is 3.9 GB, and fits within the memory budget of an iPhone 17 Pro.

How it compares

The first-party announcement is July 14, not September 19. The laptop claim fits the ternary variant and the phone claim the 1-bit variant, and the phone fit is the company's own memory budget claim, vendor-reported. The 54 gigabyte figure is not in the text read, and benchmarks were not read here. Bonsai 2, released in September, is a separate release covered in earlier notes. The model shrank, not the reasoning is the author's take.

Related work

Watch next

  • Independent runs of the phone variant.

Sources

  1. PrismML: Bonsai 27B (July 14, 2026)prismml.com

Provenance

The note above is reproduced unedited from the original post, first published on Threads on 23 September 2026 at 00:02 IST. Sources are the papers and datasets the note draws on.

View the original post
Embed this note
<iframe src="https://founder.myndlabs.tech/notes/embed/a-27-billion-parameter-model-just-moved-onto-DdmYuaIEUp-" width="480" height="420" style="border:0;max-width:100%" loading="lazy" title="A 27 billion parameter model just moved onto a laptop. prism ml's bonsai 27b, out september 19,…"></iframe>

More notes