Today, we’re announcing Bonsai 27B: the first 27B-class model to...
PrismML@PrismML
143 views
Jul 15, 2026
~3 min read
Advertisement
1
Today, we’re announcing Bonsai 27B: the first 27B-class model to run on a phone.
Bonsai 27B is the new multimodal flagship of the Bonsai family. Based on Qwen3.6 27B, it brings a new capability tier to local AI: multi-step reasoning, structured tool use, long-context workflows, and coherent agentic loops.
Until now, models in this class have been impractical to deploy locally. A 27B model occupies roughly 54 GB in 16-bit precision, and even a strong 4-bit build is around 18GB - too large for a phone and for most laptops.
Bonsai 27B changes that.
It comes in two variants:
• Ternary Bonsai 27B: 5.9 GB, 1.71 effective bits per weight, optimized for laptop-class quality.
• 1-bit Bonsai 27B: 3.9 GB, 1.125 effective bits per weight, optimized for phone-class footprint.
Everything is open-sourced today under the Apache 2.0 license.
Bonsai 27B is the new multimodal flagship of the Bonsai family. Based on Qwen3.6 27B, it brings a new capability tier to local AI: multi-step reasoning, structured tool use, long-context workflows, and coherent agentic loops.
Until now, models in this class have been impractical to deploy locally. A 27B model occupies roughly 54 GB in 16-bit precision, and even a strong 4-bit build is around 18GB - too large for a phone and for most laptops.
Bonsai 27B changes that.
It comes in two variants:
• Ternary Bonsai 27B: 5.9 GB, 1.71 effective bits per weight, optimized for laptop-class quality.
• 1-bit Bonsai 27B: 3.9 GB, 1.125 effective bits per weight, optimized for phone-class footprint.
Everything is open-sourced today under the Apache 2.0 license.
2
Raw capability determines what a model can do. Intelligence density determines where it can do it.
Bonsai 27B moves the Pareto frontier left again: 27B-class capability in a footprint smaller than many full-precision 2B models.
By intelligence density, 1-bit Bonsai 27B delivers 0.53 per GB - more than 10x the full-precision baseline and roughly 2.7x the best conventional low-bit alternative.
Bonsai 27B moves the Pareto frontier left again: 27B-class capability in a footprint smaller than many full-precision 2B models.
By intelligence density, 1-bit Bonsai 27B delivers 0.53 per GB - more than 10x the full-precision baseline and roughly 2.7x the best conventional low-bit alternative.
3
The footprint reduction does not come at the expense of the capabilities that matter.
Across 15 benchmarks spanning knowledge, reasoning, math, coding, instruction following, tool use, and vision, Ternary Bonsai 27B retains 95% of the full-precision model’s performance. The 1-bit variant retains 90%.
Math and coding remain especially strong, while tool-calling performance stays close to full precision.
Across 15 benchmarks spanning knowledge, reasoning, math, coding, instruction following, tool use, and vision, Ternary Bonsai 27B retains 95% of the full-precision model’s performance. The 1-bit variant retains 90%.
Math and coding remain especially strong, while tool-calling performance stays close to full precision.
4
Why does this matter? Because modern AI workflows are no longer single prompts. They are sustained loops.
A capable agent may take hundreds of steps: reasoning, calling tools, reading outputs, updating its state, and iterating toward a result. When every step is remote, per-token cost accumulates, latency compounds, and private files and intermediate states repeatedly cross the network.
Move those loops onto the device, and the economics change. The marginal cost of iteration collapses. Sensitive data can remain local. Agents can work offline and continue without depending on a round trip to the cloud.
That is the unlock Bonsai 27B is built for.
A capable agent may take hundreds of steps: reasoning, calling tools, reading outputs, updating its state, and iterating toward a result. When every step is remote, per-token cost accumulates, latency compounds, and private files and intermediate states repeatedly cross the network.
Move those loops onto the device, and the economics change. The marginal cost of iteration collapses. Sensitive data can remain local. Agents can work offline and continue without depending on a round trip to the cloud.
That is the unlock Bonsai 27B is built for.
5
Here is Ternary Bonsai 27B running an end-to-end agentic workflow locally with Hermes on an NVIDIA GeForce RTX 5090 GPU.
The model reasons, calls tools, reads outputs, modifies files, and surfaces insights - all on consumer hardware, while all private files, intermediate states, and iterations remain local.
The model reasons, calls tools, reads outputs, modifies files, and surfaces insights - all on consumer hardware, while all private files, intermediate states, and iterations remain local.
1:46
6
The phone threshold is even harder than the storage number suggests.
A phone exposes only part of its memory to an application, and the model must share that budget with its KV cache, activations, runtime, and the rest of the product.
At 3.9 GB, 1-bit Bonsai 27B clears that threshold with room to work, making 27B-class local AI possible on a phone for the first time.
(Demo Mode: cached & prefilled image context)
A phone exposes only part of its memory to an application, and the model must share that budget with its KV cache, activations, runtime, and the rest of the product.
At 3.9 GB, 1-bit Bonsai 27B clears that threshold with room to work, making 27B-class local AI possible on a phone for the first time.
(Demo Mode: cached & prefilled image context)
1:19
7
Links:
Blog: prismml.com/news/bonsai-27b
The Information: theinformation.com/articles/khosl…
CNBC: cnbc.com/2026/07/14/app…
Whitepaper: github.com/PrismML-Eng/Bo…
Models: huggingface.co/collections/pr…
WebGPU Demo: huggingface.co/spaces/webml-c…
API: together.ai/models/prism-m…
GitHub: github.com/PrismML-Eng/Bo…
Docs: docs.prismml.com
Discord: discord.gg/prismml
Join us: prismml.com/#careers
Blog: prismml.com/news/bonsai-27b
The Information: theinformation.com/articles/khosl…
CNBC: cnbc.com/2026/07/14/app…
Whitepaper: github.com/PrismML-Eng/Bo…
Models: huggingface.co/collections/pr…
WebGPU Demo: huggingface.co/spaces/webml-c…
API: together.ai/models/prism-m…
GitHub: github.com/PrismML-Eng/Bo…
Docs: docs.prismml.com
Discord: discord.gg/prismml
Join us: prismml.com/#careers
