Bonsai 2 27B
Bonsai 2 27B on a GPU instance in your AWS account. Apache 2.0 ternary weights derived from Qwen 3.8, served over HTTP on port 8080 by Prism ML's llama.cpp fork.
- Thinking mode on by default
- Text input, plus an optional vision projector
- 16,384-token context on this image
- Ready in about 1 minute on g5.xlarge
