Centering AI research on efficiency. discord.gg/prismml
United States
- Tweets 371
- Following 32
- Followers 28,272
- Likes 585
Today, we’re announcing Ternary Bonsai 2 27B.
Based on Qwen3.8 27B, Bonsai 2 27B is 9x smaller than its full-precision counterpart while retaining 98.2% of its aggregate benchmark performance.
Two months after the first Bonsai 27B release, the biggest change is quality. The footprint remains 5.9 GB, but the gap to full precision has narrowed materially, with particularly strong gains in agentic coding, multimodal reasoning, and long-horizon tool use.
Ternary Bonsai 2 27B is available today under Apache 2.0.
639
1,568
15,540
5,398,811
PrismML retweeted
MLX.fast is launching a 1 week contest to improve Bonsai 2 from @PrismML! Bonsai 2 is a near losslessly comrpessed version of Qwen 3.8 27b that is around 10% the size of the original. It can fit on 16gb Apple devices and even some phones.
This is a model almost anyone can run so if you weren't able to participate in the other contests, this one is for you! The community has been able to double the speed of Bonsai in under 12 hours so there is alot of headroom here.
This contest runs until October 1, have fun!
8
15
71
9,469
PrismML retweeted
The team has been hard at work putting together some best practices and demos of Bonsai 2! Give the model a shot with the recommended settings!
huggingface.co/prism-ml/Tern…
We have been very excited to see how the community has been utilizing our Bonsai 2 model, well beyond use cases that we had originally envisioned it for. We’ve spent the last few days putting Bonsai 2 27B through various examples inspired by what the community has shown us.
These examples showcase the strengths, and some of the shortcomings–particularly on longer multi-turn agentic workflows, and will be extremely helpful as we continue to improve.
We’ll be sharing a few of those demos, along with some practical guidance on the settings and prompting that get the best results from the model.
Demo repo: github.com/PrismML-Eng/Bonsa…
5
22
2,185
PrismML retweeted
its a smart model sir
We also tested Ternary Bonsai 2 27B on the 2026 IMO problems against full-precision Qwen3.8 27B (54GB) and Gemma 4 12B QAT (~7GB).
No internet. No tools. 131K-token reasoning budget.
Bonsai scored in the upper end of the human bronze-medal range and retained 95% of Qwen’s IMO score, while completing the problems in 70% of the time.
Code: github.com/PrismML-Eng/Bonsa…
5
10
175
28,733
Bonsai can build 🌱🛹
Another run, another skatepark. Add coins, add a rider, fix the controls. A few follow-ups go a long way 😄
We have been very excited to see how the community has been utilizing our Bonsai 2 model, well beyond use cases that we had originally envisioned it for. We’ve spent the last few days putting Bonsai 2 27B through various examples inspired by what the community has shown us.
These examples showcase the strengths, and some of the shortcomings–particularly on longer multi-turn agentic workflows, and will be extremely helpful as we continue to improve.
We’ll be sharing a few of those demos, along with some practical guidance on the settings and prompting that get the best results from the model.
Demo repo: github.com/PrismML-Eng/Bonsa…
1
4
12
1,331
PrismML retweeted
We asked Bonsai 2 27B to build a little pretend OS in the browser. Meet BonsaiOS 🌱
Windows, a simulated terminal, notes, and a calculator—all in one HTML file. A fun coding demo, with quirks included.
Here’s how it came together through Hermes and five feedback rounds. 🧵
Starting from an empty workspace, Bonsai 2 27B built a browser-based desktop in a single HTML file, then iteratively fixed issues across multiple rounds of feedback, from broken windows and UI behavior to complete functionality and final styling.
Code: github.com/PrismML-Eng/Bonsa…
3
4
34
4,013
We have been very excited to see how the community has been utilizing our Bonsai 2 model, well beyond use cases that we had originally envisioned it for. We’ve spent the last few days putting Bonsai 2 27B through various examples inspired by what the community has shown us.
These examples showcase the strengths, and some of the shortcomings–particularly on longer multi-turn agentic workflows, and will be extremely helpful as we continue to improve.
We’ll be sharing a few of those demos, along with some practical guidance on the settings and prompting that get the best results from the model.
Demo repo: github.com/PrismML-Eng/Bonsa…
13
10
131
14,383
Starting from an empty workspace, Bonsai 2 27B built a browser-based desktop in a single HTML file, then iteratively fixed issues across multiple rounds of feedback, from broken windows and UI behavior to complete functionality and final styling.
Code: github.com/PrismML-Eng/Bonsa…
1
1
19
6,748
PrismML retweeted
Bonsai 2 27b Computer Use showcase now available!
bonsai.puppetmaster.gg/#film…
1
5
12
1,312
PrismML retweeted
Bonsai 2 27b from @PrismML can do Computer Use ✅
Here is Bonsai 2 27b, on my own 12gb GPU, planning a trip to the Art Institute in Chicago for me.
93.5 tok/s, 56s, 6 steps with reasoning.
More details / videos : bonsai.puppetmaster.gg/
4
6
61
6,111
PrismML retweeted
The Snapdragon AR1+ and AR1 Gen 1 SoC are here, offering 4x lower memory and 2x faster throughput.
#SnapdragonSummit #Qualcomm
1
3
11
1,375
PrismML retweeted
Excited to be building the future of on-device AI with @ziad_asghar and the @Qualcomm team!
Today at Snapdragon Summit, we’re announcing 1-bit Bonsai vision-language model running locally on AI smart glasses powered by Snapdragon® AR1 Gen 1 Platform.
With Qualcomm Technologies, we optimized Bonsai for Snapdragon Hexagon™ NPU, delivering 4x lower memory usage and 2x faster token generation for equivalent intelligence, making it possible to bring significantly more capable AI to run directly on highly constrained wearable devices.
We believe model-hardware co-design will be critical to bringing increasingly capable AI from the cloud onto the devices people use every day.
1
6
12
1,384
PrismML retweeted
Am pleased to announce a 1-bit 2B Bonsai model running locally on an AI smart glasses at Qualcomm’s Snapdragon summit. Powered by Snapdragon AR1 Gen 1 this is the first time a 1-bit multi-modal model runs on a personal AI platform. And many more to follow….
Today at Snapdragon Summit, we’re announcing 1-bit Bonsai vision-language model running locally on AI smart glasses powered by Snapdragon® AR1 Gen 1 Platform.
With Qualcomm Technologies, we optimized Bonsai for Snapdragon Hexagon™ NPU, delivering 4x lower memory usage and 2x faster token generation for equivalent intelligence, making it possible to bring significantly more capable AI to run directly on highly constrained wearable devices.
We believe model-hardware co-design will be critical to bringing increasingly capable AI from the cloud onto the devices people use every day.
9
12
72
8,683
PrismML retweeted
We quantized a vision model down small enough for local inference in glasses! Realtime NPU inference for vision, translation, multilingual and more.
Today at Snapdragon Summit, we’re announcing 1-bit Bonsai vision-language model running locally on AI smart glasses powered by Snapdragon® AR1 Gen 1 Platform.
With Qualcomm Technologies, we optimized Bonsai for Snapdragon Hexagon™ NPU, delivering 4x lower memory usage and 2x faster token generation for equivalent intelligence, making it possible to bring significantly more capable AI to run directly on highly constrained wearable devices.
We believe model-hardware co-design will be critical to bringing increasingly capable AI from the cloud onto the devices people use every day.
5
24
1,490
PrismML retweeted
Making smart glasses smarter at @PrismML
Today at Snapdragon Summit, we’re announcing 1-bit Bonsai vision-language model running locally on AI smart glasses powered by Snapdragon® AR1 Gen 1 Platform.
With Qualcomm Technologies, we optimized Bonsai for Snapdragon Hexagon™ NPU, delivering 4x lower memory usage and 2x faster token generation for equivalent intelligence, making it possible to bring significantly more capable AI to run directly on highly constrained wearable devices.
We believe model-hardware co-design will be critical to bringing increasingly capable AI from the cloud onto the devices people use every day.
5
13
866