# Bonsai

## Docs

- [Formats & Runtime Support](https://docs.prismml.com/download/formats.md): GGUF vs MLX, the Q1_0 and Q2_0 quantization types, and which backends have native kernels.
- [Download the Models](https://docs.prismml.com/download/models.md): Hugging Face repositories, file names, and download commands for every Bonsai variant.
- [Introduction](https://docs.prismml.com/get-started/introduction.md): Bonsai is a family of 1-bit and ternary language models from PrismML.
- [Quickstart](https://docs.prismml.com/get-started/quickstart.md): Clone the demo repo, run one setup command, and chat with Bonsai locally.
- [Try It Without Installing](https://docs.prismml.com/get-started/try-online.md): Hosted GPU demo, in-browser WebGPU inference, a Colab notebook, and iOS apps.
- [Hermes](https://docs.prismml.com/integrations/hermes.md): Run Nous Research's Hermes terminal agent on local Bonsai.
- [openclaw](https://docs.prismml.com/integrations/openclaw.md)
- [Bonsai 1.7B](https://docs.prismml.com/models/bonsai-1-7b.md): The smallest Bonsai: 0.25 GB on disk for the tightest memory budgets.
- [Bonsai 27B](https://docs.prismml.com/models/bonsai-27b.md): The flagship Bonsai: vision-language, native tool calling, and a 262K-token context, in binary or ternary weights.
- [Bonsai 4B](https://docs.prismml.com/models/bonsai-4b.md): Half the footprint of 8B for edge deployments
- [Bonsai 8B](https://docs.prismml.com/models/bonsai-8b.md): 8.2B parameters in 1.16 GB — runs on laptops and phones at up to 368 tok/s.
- [Community & Resources](https://docs.prismml.com/resources/community.md): Whitepapers, source repositories, community benchmarks, and where to get help.
- [Troubleshooting](https://docs.prismml.com/resources/troubleshooting.md): Fixes for the failure modes people actually hit: kernels, memory, ports, and platform quirks.
- [llama.cpp](https://docs.prismml.com/run/llamacpp.md): Run Bonsai GGUF with llama.cpp: pre-built binaries, CLI usage, and building from source.
- [MLX](https://docs.prismml.com/run/mlx.md): Run Bonsai MLX weights on Apple Silicon for the best Mac performance.
- [open-webui](https://docs.prismml.com/run/open-webui.md)
- [Run the Server](https://docs.prismml.com/run/server.md): Start Bonsai's OpenAI-compatible server and call it from curl, Python, or any client.