GitHub radar
NeoHorse-1-9B: a laptop-friendly 9B agent model
NeoHorse-1-9B is a post-trained model from TokenRhythm based on Qwen3.5-9B that is optimized for agent flow, tooling, and programming. It is also quantized to fit on your 16GB laptop!
NeoHorse-1-9B is our post-trained model on top of Qwen3.5-9B optimized for use in agent flows. This model can call tools and program while following complex instructions. On ten benchmarks, it achieves a macro avg score of 69.04 vs 65.60 for Qwen3.5-9B. With a native context length of 262,144 tokens, you can fit an entire small project into a single session. We provide quantized GGUF weights for use with Ollama or LM Studio. At 4-bit quantization, this model requires around ~5.4 GB of memory and will run on a laptop with 16 GB of RAM. While this model isn’t suited for cutting edge agent flows, it is a good representation of what is possible with a model this size.
Why a vibe-coder should care
This model can easily be deployed on a normal laptop for running local agents to do simple coding or automation. With a 262k context length, you can easily put an entire small project into a single session without having to break it up. This model represents a good representation of what is possible with a 9B model. It is not as good as Claude or GPT, but it will get the job done if you need to do some offline repetitive tasks.
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Install the NeoHorse-1-9B model from TokenRhythm locally: https://huggingface.co/TokenRhythm/NeoHorse-1-9B — find the right GGUF version for my machine, install via Ollama, and show me how to use it for coding tasks.
A regular laptop with 16 GB of RAM — the 4-bit version takes approximately 5.4 GB.
Open on Hugging Face▌ More finds