Needle2

AI

small on-device tool-calling model

SHOW HN2026

537points
185comments
#42of 4,366 in 2026
Top 1%of 2026 by points

How HN took it

CuriousMostly questions and neutral comments

18 positive, 57 neutral or asking, 4 mixed, 29 critical, out of 108 comments about the project. Thread score -0.07.

Performance: 5 for, 16 againstIdea: 9 for, 4 againstExecution: 1 for, 7 againstDocs: 0 for, 1 againstPrivacy: 0 for, 1 against

The thread, comment by comment

Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.

positive neutral or mixed critical about another commenter or something else the maker

What HN said

What people liked

  • Impressively small model that runs locally in WASM

    From 2 comments
    • Its pretty significant you've got this working locally in wasm. Very cool. Re: robotics: I'm unsure how this could be helpful. It fails a pretty simple navigation prompt. X0:…
      kooi on HN
    • This is cool! While most of the industry focuses on the frontier of “intelligence” (function), a release like this represents the frontier of the other end of the spectrum…
      redrix on HN
  • Fills a useful niche between non-LLM tools and frontier models

    From 2 comments
    • Utterly Fascinating! For the longest time, I conceptualized LLM's as Text Input -> Text Output transformers, then later as Text Input -> Video Output transformers. Later still I…
      peter_d_sherman on HN
    • Looks incredibly cool, although it ran at ~4 tok/s on my browser (I assume because I've disabled V8). I believe your numbers though. I always wanted to try using something like…
      a022311 on HN

Objections

  • Gives wrong tool calls for simple commands like thermostat and lights

    From 3 comments
    • My first query: > Make it a little warmer in here. The reply: > "name": "set_thermostat", > "arguments": { > "temperature": 65, > "mode": "cool", > ... > "reasoning": "'warmer'…
      dbeardsl on HN
    • > turn on the tv { "function_calls": [ { "name": "lock_door", "arguments": { "door": "tv" } } ], "confidence": 0.0158 } Very interesting, seems confidence is 0 when tool calls…
      dangoodmanUT on HN
    • > Turn the lights down low in the bedroom Sets lights to 30% but also off > Turn the lights low in the bedroom Sets lights to on This is a cool idea but I think humans assume…
      Robin_Message on HN
  • Confidence scores are unreliable, sometimes high for wrong answers

    From 2 comments
    • Specifying units seems to be unreliable; I tried adding a description to the set_thermostat temperature: ``` "temperature": { "type": "number", "description": "degrees…
      wky on HN
    • > set lights on { "function_calls": [ { "name": "set_lights", "arguments": { "room": "bedroom", "state": "on" } } ], "confidence": 0 } is confidence 0 that its confident?
      tough on HN
  • Fails on arithmetic and units without clearer tool descriptions

    From 2 comments
    • I tested with ``` import needle @needle.tool def add(a: int, b: int): "Add two numbers." return a + b agent = needle.Needle(tools=[add]) print(agent.run("calculate 1 +…
      hathym on HN
    • It seems to fail. I sent the prompt: “ 5° warmer” And it said: “ setting the temperature to 5°F”
      raylad on HN

Requests

  • Ollama release

    From 1 comment
  • ESP32 and desktop x86 support

    From 2 comments
    • Any instructions available for running this on an ESP32-S3 or P4 like the site says?
      forsalebypwner on HN
    • I tried using it on my x86_64 desktop. Could not use it. Then I tried using it on a raspberry pi 4. Did not work. This is specifically for 64 bit arm. I see no reason why you…
      israrkhan on HN
  • Home Assistant plugin

    From 1 comment
    • If you want Needle2 to rget lots of testing, become well known, etc - make a Home Assistant plugin.
      KennyBlanken on HN

Questions

  • Why 2-bit quantization and 14MB size choice

    From 2 comments
    • This is cool. I definitely think the "micro" sized LLM space is underappreciated, so it's always good to see work like this. I foresee a paradigm in some contexts where you have…
      nater5000 on HN
    • Curious, why did you go down to 2bit rather than 4 bits? 4bit with folding the layers should arrive at the same size, but with better quality?
      r0ze-at-hn on HN
  • How it differs from LiteRT

    From 1 comment
    • Congratulations. 28MB is impressive, I've not played around with actual queries/outputs. I'm wondering what is the overall thesis/plan here and where exactly the innovation lies?…
      havercosine on HN

Alternatives named

  • LiteRT / TensorFlow Lite

    From 1 comment
    • Congratulations. 28MB is impressive, I've not played around with actual queries/outputs. I'm wondering what is the overall thesis/plan here and where exactly the innovation lies?…
      havercosine on HN
  • FunctionGemma 270M

    From 1 comment
    • That's really cool - I was already thinking of compressing `functiongemma-270m-it` down to 1-2 bits so it would work flawlessly in the browser. Your `Fine-tuning` feature is even…
      arthuqa on HN
  • LFM2.5 230M and Apple Foundation Model

    From 1 comment

Category, keywords, summaries and labels on this page were made by AI (Thread Reader v2) and not checked by hand. Each point links to the comments it came from.

Add a badge to your READMEShown on HackerShow
[![Shown on HackerShow](https://hackershow.org/badge.svg)](https://hackershow.org/show/49246804)