LiquidAI has released LFM2_5 2.6B, an agentic model designed to operate entirely on-device. This model supports planning, tool calling, and multi-step tasks without requiring cloud API access. It demonstrates varying performance speeds across different hardware, decoding 220 tokens/s on an M5 Max CPU and 113 tokens/s on a Ryzen AI Max+ 395, while using under 2.5 GB of memory. AI
IMPACT Enables on-device AI capabilities for planning and multi-step tasks, potentially reducing reliance on cloud infrastructure.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Mastodon — fosstodon.org →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
