Edge Model LFM2.5 Matches DeepSeek-V4 in Tool Calling at 3.7x the Speed
maximelabonne · x · 2026-08-07
For simple tool calling, edge models now demonstrate impressive capabilities. Liquid AI's LFM2.5-2.6B recently matched DeepSeek-V4 level performance in practical testing.
Running on a single machine with 4x RTX 5090s, both models successfully completed a complex job requiring 35 tool calls. LFM2.5 achieved a speed of 366 tok/s, finishing in just 19 seconds, whereas DeepSeek-V4-Flash ran at 77 tok/s and took 70 seconds. This highlights the potential of purpose-built small edge models for agentic workflows, capable of running directly on phones.
Related event: Liquid AI 2.6B Model Matches DeepSeek in Tool Calling, 3.7x Faster(3 posts)→
More from coding & agent
- ERP•AI launches headless SaaS with $0 seats that bills machines: $1 buys 4,000 record writes — rohanpaul_ai · 2026-10-16
- FDE-Gym benchmark shows huge gap on real AI delivery tasks: top model scores 0.563 — DeepLethe · 2026-10-16
- Vibe-coded dashboard leaked franchise financials before anyone logged in — shauntrennery · 2026-10-16
- Fintech HR built a sourcing agent on 60,000 past applicants — 30 qualified candidates in a month — G0NZ0BOY · 2026-10-16
- Orchestrating a 100-image portrait of humanity for aliens with ChatGPT, Argon and p5.brush — dx8xb · 2026-10-16
- 3x DGX Spark cluster runs GLM-5.3-Flash locally behind a proxy, 1.75M tokens/day at 119W — Teknium · 2026-10-16