GitHub radar
Ornith-1.5-9B: Vision and Agents on a Laptop
Small multimodal model by ornith-ai. Can process images, call tools, perform agentic actions, runs on a 16 GB laptop.
Ornith-1.5-9B is a 9.7 billion parameter dense language model developed by ornith-ai on top of Qwen3.5 and Gemma4, further pretrained and trained with RL. What sets this model apart is its self-improvement training paradigm where the model was able to generate tasks for itself, solve them, and improve without needing static datasets from humans. This model can take both text and image inputs, call tools, and operate within agentic workflows like those powered by MCP. When used in an agentic workflow using OpenHands harness, it achieves a score of 70.6 on SWE-bench Verified which is quite good for a 9B model. You can find a quantized GGUF version of this model on Hugging Face, which runs with ~6GB of memory at 4-bit.
Why a vibe-coder should care
If you’re looking for a local agent and a model capable of processing both text and images without requiring a cloud subscription, Ornith-1.5-9B should work fine on your 16 GB laptop. It’s great for providing assistance during coding and other tasks that require both textual and visual inputs.
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Install Ornith-1.5-9B locally: https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF — pick the right quantized variant for my Mac, run it via Ollama, and show me how to send an image along with a question
A laptop with 16 GB of RAM or more — the 4-bit version takes around 6 GB.
Open on Hugging Face▌ More finds