Agents / Liquid AI

LFM2.5-24B-A2B

Liquid AI’s largest LFM2.5 model: a 24B mixture-of-experts reasoning model with about 2.45B active parameters per token, built for fast private agents and tool calling on your own computer.

Verified sourceReleased Oct 13, 2026Source checked 11/9/2026

LOCALRENTED GPUOPEN WEIGHTS

Freshness 88 out of 100. Release age: 32 days old, Oct 13, 2026. Measures age, not quality.

Popularity 63 out of 100. 35,825 downloads in 30 days; 613 lifetime likes. Checked Nov 13, 2026. Measures adoption, not quality.

What do these scores mean?

Two ways to compare models. Neither score measures answer quality, benchmark performance, or whether a model fits your machine.

Freshness · release age

A new release starts at 100. The score halves every 180 days: 100 × 2^(−age in days / 180). When the release date is unavailable, a repository date is a fallback—not proof of when the model was released. Missing dates stay unscored.

Check the date source

Popularity · adoption

Downloads over the past 30 days carry 80% of the score; lifetime likes carry 20%. Both use a logarithmic scale, capped at 10 million downloads and 100,000 likes. If likes are unavailable, downloads carry all the weight. Missing download statistics stay unscored.

round(80 × min(1, log10(1 + downloads) / 7) + 20 × min(1, log10(1 + likes) / 5))

Check the statistics source

Scores as of Nov 14, 2026. Statistics checked Nov 13, 2026. Statistics are marked for refresh after 72 hours.Full scoring method

WORKFLOWS FOR THIS MODEL 3

What do you want to do?

Choose a task. Get the software, files, settings, and steps to run it.

MAKE IT PERSONALChoose your hardware

Private chat

Chat privately with LFM2.5-24B-A2B

Run Liquid AI’s LFM2.5-24B-A2B with llama.cpp on Windows, Mac or Linux and chat in your browser, with every prompt staying on your computer.

Freshness 88 out of 100. Release age: 32 days old, Oct 13, 2026. Measures age, not quality.

llama.cpp

2 packagesmacOS / Windows / Linux

Open chat guide

AI agent

Build a local AI agent with LFM2.5-24B-A2B

Run LFM2.5-24B-A2B with llama.cpp, connect Hermes Agent using Liquid AI’s published settings, and have it read two files and write a checked review.

Freshness 88 out of 100. Release age: 32 days old, Oct 13, 2026. Measures age, not quality.

Hermes Agentllama.cpp

2 packagesmacOS / Windows / Linux

Open agent guide

Coding

Code locally with LFM2.5-24B-A2B in VS Code + Cline

Serve LFM2.5-24B-A2B with llama.cpp, connect VS Code and Cline, and have it write a small module plus tests that you review and run yourself.

Freshness 88 out of 100. Release age: 32 days old, Oct 13, 2026. Measures age, not quality.

VS Code + Clinellama.cpp

2 packagesmacOS / Windows / Linux

Open coding guide

Open a guide for its software, hardware requirements, and setup steps.

At a glance

Parameters: 24.0B total · 2.45B active
Architecture: Hybrid MoE · 30 gated short-convolution + 10 GQA layers, 64 experts (top-4)
Context length: 131,072 native
License: LFM Open License v1.0 · commercial use allowed with conditions; see the license
Software: llama.cpp, MLX, vLLM, SGLang, Transformers

MY HARDWARE

Will it run on your machine?

Save your machine to see a personalized rating and its reasoning.

Add my hardware

Best for

Run a local Hermes agent that reads and writes files with your approval.
Connect VS Code + Cline for private coding help.
Private reasoning and planning chat in your browser.

Tradeoffs

Always reasons before answering; there is no non-thinking mode.
Liquid AI says it is not recommended for knowledge-heavy tasks on its own; pair it with retrieval or search.
Text only.
Trails Qwen3.6-35B-A3B on several publisher-reported benchmarks, including SWE-Bench Verified (51.0 vs 66.2).

Ways to run it

llama.cpp · Windows, macOS, Linux

Publisher GGUF. Used by the chat, Cline and Hermes guides on this page.

llama-server -hf LiquidAI/LFM2.5-24B-A2B-GGUF:Q4_K_M --jinja -c 32768

Official instructions

MLX · Apple Silicon

Liquid AI’s 13.6 GB MLX 4-bit build for mlx-lm.

Official instructions

vLLM / SGLang · Linux GPU server

BF16 or the publisher FP8 checkpoint for GPU serving.

vllm serve LiquidAI/LFM2.5-24B-A2B

Official instructions

What to expect

Start with the Q4_K_M GGUF in llama.cpp. It is the most practical local agent model in its size class when you have 24 GB of GPU memory or a 32 GB Mac.

Explore its uses

AGENTSTOOL CALLINGREASONINGCODINGMIXTURE OF EXPERTSGGUFMLX

REFERENCE LIBRARY

Sources and files

Primary sources and files behind this model page.

4 SOURCES

01Liquid AI model cardhuggingface.co 02Official GGUF fileshuggingface.co 03Official MLX 4-bit buildhuggingface.co 04LFM Open License v1.0huggingface.co

Keep exploring

Agents

Z.ai

GLM 5.3

Explore 2 versions of GLM 5.3. Compare model sizes, packages, hardware requirements, and available setups.

2 versionsCompare sizes & formats

Kimi K3

An open-weight multimodal model for long-running coding, reasoning, and knowledge work.

Freshness 55 out of 100. Repository age: 154 days old, Jun 13, 2026. Measures age, not quality.

Arena Open Source Agents: rank 5 of 21. Ranked model: Kimi K3 (Max). Export dated Nov 11, 2026. Source checked Nov 13, 2026. Human preference ranking, not hardware suitability.+2

MODEL SIZE2.8T total · 104B active

LOCALRENTED GPU

Repository opened Jun 13, 2026

Agents

Google DeepMind

Gemma 4

Explore 4 versions of Gemma 4. Compare model sizes, packages, hardware requirements, and available setups.

4 versionsCompare sizes & formats

LOCALRENTED GPU

Compare versions →