PulseAugur
EN
LIVE 07:31:46
Français(FR) Liquid # AI publie LFM2.5-2.6B : 2,6 milliards de paramètres, pensé pour tourner entièrement sur l’appareil. Les données ne quittent jamais le terminal, coût ma

Liquid AI releases on-device agentic model LFM2.5-2.6B with 128K context

By PulseAugur Editorial · [4 sources] ·

Liquid AI has released LFM2.5-2.6B, an open-weights, on-device agentic model designed for mobile and edge devices. This model boasts 2.69 billion parameters, a 128,000-token context window, and can perform multi-step tasks, including planning and tool calling, without relying on cloud APIs. Liquid AI reports that LFM2.5-2.6B achieves competitive performance on instruction-following and tool-use benchmarks compared to models significantly larger than itself, while maintaining low memory usage and fast decoding speeds on various hardware. AI

IMPACT Enables advanced agentic capabilities on edge devices, reducing reliance on cloud infrastructure and potentially lowering costs for on-device AI applications.

RANK_REASON Frontier-lab model release with open weights and system card.

Read on Mastodon — fosstodon.org →

AI-generated summary · Google Gemini · from 4 sources. How we write summaries →

Liquid AI releases on-device agentic model LFM2.5-2.6B with 128K context

COVERAGE [4]

  1. MarkTechPost TIER_1 English(EN) · Asif Razzaq ·

    Liquid AI Releases LFM2.5-2.6B: An On-Device Agentic Model With 128K Context, Tool Calling, And Open Weights

    <p>Liquid AI released LFM2.5-2.6B, an agentic model that plans, calls tools, and completes multi-step tasks entirely on-device. The 2.69B parameter model pairs 22 double-gated short convolution blocks with 8 GQA blocks across 30 layers, handles 131,072 tokens of context, and deco…

  2. dev.to — LLM tag TIER_1 English(EN) · Prabhakar Chaudhary ·

    LFM2.5-2.6B: How Liquid AI Built an Agentic Model That Runs on Your Phone

    <h1> LFM2.5-2.6B: How Liquid AI Built an Agentic Model That Runs on Your Phone </h1> <p>The promise of AI agents running entirely on-device — no cloud calls, no per-token billing, no latency from a round-trip to a remote server — has been around for a while. The problem is that m…

  3. Mastodon — fosstodon.org TIER_1 Français(FR) · [email protected] ·

    Liquid AI releases LFM2.5-2.6B: 2.6 billion parameters, designed to run entirely on-device. Data never leaves the terminal, cost ma

    Liquid # AI publie LFM2.5-2.6B : 2,6 milliards de paramètres, pensé pour tourner entièrement sur l’appareil. Les données ne quittent jamais le terminal, coût marginal proche de zéro. Sa dernière étape de post-entraînement : du RL agentique multi-tours dans # Hermes Agent, Pi et #…

    LINKS github.com/…/2026-08-05-hermes-agent-news…
  4. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    🤖 # LiquidAI released # LFM2_5 2.6B, an # agentic model that runs entirely on-device: planning, tool calling & multi-step tasks without any cloud API # AI # LLM

    🤖 # LiquidAI released # LFM2_5 2.6B, an # agentic model that runs entirely on-device: planning, tool calling & multi-step tasks without any cloud API # AI # LLM # EdgeAI # opensource 🧵👇 ⚡ Decodes 220 tokens/s on an M5 Max CPU, 113 tokens/s on a Ryzen AI Max+ 395 and 30 tokens/s o…

RELATED ENTITIES

RELATED TOPICS