LaunchAI ModelsAugust 14, 2026

Cactus Compute releases Needle 2, a 45M-parameter tool-calling model

Needle 2 ships as a 14MB binary and runs a full agent session in ~28MB of RAM. It trades wins with FunctionGemma 270M, LFM2.5 230M, and Apple FM at 5 to 70x smaller.

How this story unfolded

3 days · 2 reports · 3 community posts · from Aug 10

  1. Aug 10
  2. Aug 12
  3. Aug 13
  4. Aug 14

Daily brief

Get tomorrow's AI brief in your inbox

More stories today

Open the live feed