PrismML's Ternary Bonsai 2 27B is 9x smaller at 5.9GB while keeping 98.2% benchmark performance
alexcovo_eth · x · 2026-09-19
PrismML released Ternary Bonsai 2 27B, a ternary-quantized version of Qwen3.8 27B that is 9x smaller at 5.9GB while retaining 98.2% of the full-precision model's aggregate benchmark performance.
The biggest change over the first Bonsai 27B released two months ago is quality: same footprint, but a materially narrower gap to full precision, with strong gains in agentic coding, multimodal reasoning, and long-horizon tool use. Available today under Apache 2.0 and runs locally.
Related event: PrismML's Ternary Bonsai 2 27B Shrinks to 5.9GB, Retains 98.2% Performance(20 posts)→
More from Infra
- Training-free page-sparse decode in llama.cpp: 1.72× faster 128k decoding on an RTX 5090 — topppggg · 2026-11-16
- llama.cpp offloads CI caches and runners to Hugging Face to survive 35 PRs/day — ggerganov · 2026-11-16
- India's Sarvam launches Token Factory, hosting top open models on local infra — HarveenChadha · 2026-11-16
- ZGX Fury stress-tested: 351 tok/s on Qwen3.8-Flash-Next, offline agents and local RAG demos — Sam Witteveen · 2026-11-16
- Gemini 3.5 Flash now runs on air-gapped Google Distributed Cloud, offline up to 12 months — shashib · 2026-11-16
- Gemma.c: full Gemma 1/2/3 multimodal inference in a single C file, ~72 tok/s on CPU at INT8 — This-Purchase-3325 · 2026-11-16