model

Bonsai 2 27B

Ternary-weight, Apache 2.0-licensed derivative of Qwen3.8 27B distributed by PrismML in GGUF and MLX packagings with an optional 8-bit vision component.

Known aliases

Bonsai 2

Relationships

No evidence-backed relationships are recorded.

Current clusters

product1 publisher

PrismML shrinks Qwen3.8 27B model to 5.9GB by storing each weight as +1, -1 or zero

Bonsai 2 27B keeps 98 percent of Qwen3.8 27B's aggregate benchmark score, up from 95 percent for PrismML's first release, and TechCrunch reports the 5.9GB file is small enough for a PC and possibly a high-end phone.

Publishers:techcrunch.com

Reality

Evidence40
Adoption42
Hype gap+30
Incentives72
Confidence45

build1 publisher

Loading PrismML's 5.95 GB Bonsai 2 requires the company's own llama.cpp fork

PrismML's ternary build of Qwen3.8 27B keeps 98.2 percent of the full-precision benchmark average on both of the company's inconsistent scorecards, and the loss it does take is concentrated in knowledge and reasoning.

Publishers:runtimewire.com

Reality

Evidence42
Adoption14
Hype gap+30
Incentives74
Confidence52