LFM2.5-24B-A2B

Name: LFM2.5-24B-A2B
Author: Liquid Ai

Liquid Ai🇺🇸 United States

active

Compare with other models →

Context window131K tokens

Version History

LFM2.5-24B-A2BminorOctober 13, 2026

LFM2.5-24B-A2B extends LFM2-24B-A2B with a 131K-token context window (up from 32K), a 128K vocabulary (up from 65K), and a reasoning-only design. Pre-training grew from ~17T to ~33T tokens, with agentic post-training in harnesses such as OpenClaw, Hermes Agent, Pi, and OpenCode.

Coverage

model releaseLiquid Ai

Liquid AI releases LFM2.5-24B-A2B: open-weight 24B reasoning MoE with 2B active parameters and 131K context

Liquid AI released LFM2.5-24B-A2B, an open-weight reasoning mixture-of-experts model with 24B total and 2B active parameters and a 131K-token context window. The company claims it is competitive with Qwen3.6-35B-A3B on agentic and tool-use tasks while decoding at 184 tokens/s on an Apple M5 Max.

October 13, 2026 · 5:20 PM3 min read

Liquid AI LFM2.5 open-weight