LaunchAI ModelsAugust 19, 2026

Liquid AI releases LFM2.5-2.6B on-device agentic model

LFM2.5-2.6B has 2.69B parameters, a 131,072-token context window, and open weights, designed for on-device tool-calling. Community tests report 17-30 tok/s on phones via Q4_K_M GGUF, with benchmarks competitive against models 4x larger.

How this story unfolded

2 weeks · 4 reports · 7 community posts · 11 of 12 shown

Aug 4
- LiquidAI/LFM2.5-2.6Bhuggingface.co
Aug 5
- LFM2.5-2.6B on a OnePlus 13 at 17 tok/s ~ Pure CPU
Aug 6
- No cloud, no GPUs, no problem: Liquid AI's new model LFM2.5-2.6B brings powerful AI agents to devices as small as a Raspberry Piventurebeat.com
Aug 7
- Liquid AI Releases LFM2.5-2.6B: An On-Device Agentic Model With 128K Context, Tool Calling, And Open Weightsmarktechpost.com
- LFM2.5-2.6B model+KV cache quantization report
Aug 14
- LFM 2.5 2.6B is the best small model for tool use I have ever used.
Aug 19
- LFM2.5 Q4\_0 Checkpoints from Quantization-Aware Distillationhuggingface.co
- LFM 2.5 QAD

Daily brief

Get tomorrow's AI brief in your inbox

Liquid AI releases LFM2.5-2.6B on-device agentic model

How this story unfolded

Daily brief

More stories today