Cactus has released Needle 2, a 14MB agentic LLM designed for resource-constrained devices like phones, wearables, and robots. This compact model operates with 28MB of RAM and achieves impressive speeds, running at 500 tokens/sec on a Raspberry Pi 5 and hundreds of tokens/sec on smartphones. Needle 2 utilizes Simple Attention Networks, making it significantly more efficient than conventional transformers, and offers capabilities for tool calling and structured data extraction. AI
IMPACT Enables more powerful AI capabilities on low-power, resource-constrained devices, expanding the reach of AI beyond traditional computing platforms.
RANK_REASON This is a product release for a specialized LLM targeting edge devices, not a frontier model release from a major lab.
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
AI-generated summary · Google Gemini · from 3 sources. How we write summaries →
