Cactus has released Needle 2, a compact 14MB agentic LLM designed for on-device AI applications. This model, featuring 45 million parameters compressed to 2-bit precision, operates efficiently on low-power hardware like smartphones, wearables, and microcontrollers, requiring as little as 28MB of RAM. Needle 2 demonstrates competitive performance in tool-calling and structured extraction tasks, rivaling larger models while consuming significantly less power and memory. The model is open-source under the Apache 2.0 license, with weights available on Hugging Face and a dependency-free C++ engine for easy deployment. AI
IMPACT Enables advanced AI capabilities on resource-constrained devices, potentially expanding the reach of AI into IoT and wearables.
RANK_REASON Release of a specialized, small-footprint LLM for edge devices.
Read on Hacker News — AI stories ≥50 points →
- Apple Foundation Model
- Apple Vision Pro
- Cactus
- LFM2.5 230M
- Meta Quest 3S
- Needle 2
- Raspberry Pi 5
- Samsung A-Series
- Mastodon
- Apache 2.0
- Apple FM
- ESP32-S3
- FunctionGemma 270M
- Hugging Face
- Simple Attention Network
AI-generated summary · Google Gemini · from 4 sources. How we write summaries →