A 45M-parameter tool-calling model in a single 14 MB binary that runs a session in about 28 MB of RAM.
pip install cactus-needleNeedle 2 is an open model for tool calling, device use and structured extraction on tiny devices: phones, wearables, smart home gear and robots. The weights are baked into one small binary.
pip install cactus-needle
Describe your tools, then call them from Python. Extras add LoRA fine-tuning.
Nous Research · Agent
Nous Research's self-improving agent, which creates skills from experience, improves them in use and remembers you across sessions.
thedotmack · Plugin
Persistent memory for coding agents. It records what the agent did in each session and feeds the useful parts into the next one.
Stably AI · Tool / CLI
A desktop workspace for running a fleet of coding agents in parallel, each in its own git worktree, tracked in one place.