PrismML releases Ternary Bonsai 2 27B: under 6GB with 98.2% of FP16 performance
On September 18, PrismML released Ternary Bonsai 2 (27B) on Hugging Face, a ternary-quantized inference model based on Qwen3.8-27B. The model quantizes weights to the ternary values -1/0/1, compressing its size to 5.95GB (versus 54GB for the FP16 original, roughly a 9x reduction) while retaining 98.2% of aggregate benchmark performance, and can run directly in a WebGPU browser.
Confirmed
- The model is based on Qwen3.8-27B's hybrid attention architecture, unchanged except for the ternary quantization
- Weight size is 5.95GB, under 6GB, about 1/9 of FP16
- Benchmarks reach 98.2% of Qwen3.8 27B FP16 scores
- Tested running on an M5 Max, supporting in-browser WebGPU inference
- Released two months after the first-generation Bonsai 27B, with the biggest improvement in post-quantization quality; memory footprint remains around 5.9GB
- The model targets agentic use cases, and the collection is available on Hugging Face
Why it matters
Ternary quantization compresses weights to -1/0/1, dramatically reducing memory footprint and inference cost, and is a representative technique for on-device and low-cost deployment. Multiple sharers noted that a 6GB-class size lets a 27B-parameter model run on consumer devices or even in a browser—if the quality loss is controllable, it could change how large models are distributed and used.
2026-09-18 ~ 2026-09-18 · 6 related posts
Primary sources
- Ternary Bonsai 2: 27B Model Under 6GB Runs In-Browser on WebGPU, Keeps 98.2% Quality — xenovatech ·
- Prism ML releases ternary Bonsai 2 collection, including a 27B model — No-Name-Person111 ·
- Ternary Bonsai 2 27B: 9x smaller, 98.2% of full-precision performance, Apache 2.0 — cephaloform · 2026-09-18
- [source] Ternary Bonsai 2: 27B Model Under 6GB Runs In-Browser on WebGPU, Keeps 98.2% Quality — xenovatech · 2026-09-18
- Ternary Bonsai 2 hits Hugging Face: 27B ternary reasoning model under 6GB, runs in browser — cephaloform · 2026-09-18
- [source] Prism ML releases ternary Bonsai 2 collection, including a 27B model — No-Name-Person111 · 2026-09-18
- Bonsai 2 27B ships with ternary weights: 5.95GB model hits 98.2% of FP16 benchmarks — airesearch12 · 2026-09-18
1 near-duplicate retellings: cephaloform