Executive TL;DR:

• Needle2 is a 14MB agentic LLM designed for phones, wearables, and smart home devices.
• The model has sparked debate among experts about its potential and limitations.
• Some critics question the model’s small size and inconsistent performance.

The Buzz Score

The Internet’s Verdict: 60% Hyped, 40% Skeptical

Expert Reactions

Experts are divided on the potential of Needle2. Some see it as a promising development in the field of micro LLMs.

This is cool. I definitely think the ‘micro’ sized LLM space is underappreciated, so it’s always good to see work like this.

Others are more skeptical, questioning the model’s small size and performance.

Funny result from the web demo. I’m well aware that it’s an extremely small and, well, stupid, model, but even so: Query: HN Result: { ‘function_calls’: [ { ‘name’: ‘lock_door’, ‘arguments’: { ‘door’: ‘front door’ } } ], ‘reasoning’: ‘User wants to lock the door. No specific door mentioned, so use ‘front door’ as default.’, ‘confidence’: 0 }

Technical Details

The creators of Needle2 have not disclosed the exact process of creating the model, but some experts speculate that it may involve whittling down a larger model like DeepSeek.

Could someone please share how such open source micro-LLMs might have been created? Do the creators take something like DeepSeek, and then delete most of the neurons to whittle down the size?

Focus Keyword: Micro LLM