Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots

Practical AI: Tools, Models & Frameworksllmfoundation modelfine-tuningtransformer

What changed

Count budget phones, Raspberry Pis, microcontrollers, wearables, small robots like Reachy Mini, and connected home devices, and roughly four in five edge devices cost under $200. That is the hardware Needle targets: no GPU, no NPU, a few hundred MB of RAM. Baselines run the released checkpoints under vLLM at full context, and Apple FM runs on-device.

Why it matters

A concrete addition to Practical AI: Tools, Models & Frameworks: it changes what's available to builders today rather than being general commentary.

How it compares

Related prior coverage to compare against:

  • Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows
  • A $500 RL fine-tune of a 9B open model beat frontier models on catalog review
  • Show HN: Distilling DeepSeek into GPT-OSS doesn't transfer censorship. Try it

Sources