LFM2.5-24B-A2B

FP8 · vllm · 128K context · RTX PRO 6000 Blackwell · 96 GB

Liquid AI's 24B hybrid mixture-of-experts (about 2B active; 30 short-convolution and 10 attention layers, 64 experts). It reasons before it answers and calls tools.

272 tok/sdecode

12,251 tok/sprefill

128Kcontext window

Tested on this cardOct 30, 2026

Run it

Follow the steps below to download and run this recipe. Omarchy Local AI lists the recipes it can launch on your hardware. The server then answers on http://localhost:8000/v1.

1 Download the weights

hf download LiquidAI/LFM2.5-24B-A2B-FP8 \
  --revision f54bf39547fd5b7955abb9fe8d1b151e31fbbddb \
  --local-dir ~/models/LFM2.5-24B-A2B-FP8-f54bf395

2 Start the server

docker run --rm \
  --gpus all \
  -p 8000:8000 \
  --shm-size 16g \
  -v ~/models/LFM2.5-24B-A2B-FP8-f54bf395:/models/LFM2.5-24B-A2B-FP8-f54bf395:ro \
  --entrypoint vllm \
  vllm/vllm-openai:v0.30.0@sha256:8a69ffad015f138d7170c4ddc429e230a3bc1c1719f67e14324749df200a4b90 \
  serve \
  /models/LFM2.5-24B-A2B-FP8-f54bf395 \
  --served-model-name LFM2.5-24B-A2B-FP8-f54bf395 \
  --host 0.0.0.0 \
  --port 8000 \
  --max-model-len 131072 \
  --max-num-seqs 1 \
  --gpu-memory-utilization 0.9 \
  --kv-cache-dtype auto \
  --enable-auto-tool-choice \
  --tool-call-parser lfm2 \
  --reasoning-parser deepseek_r1

What it passed

✓ loadLoads and serves within an hour

✓ chatAnswers a plain question and stops on its own

✓ reasoningThinks separately and gets 17 × 23 right

✓ toolsCalls a tool with the right arguments and uses the result

✓ contextRecalls a code buried in a prompt that fills 85% of the window

✓ speedDecodes at 15 tok/s or more

Passed all six checks on a real RTX PRO 6000 Blackwell WS (450 W cap) on Oct 30, 2026.

Details

Weights: LiquidAI/LFM2.5-24B-A2B-FP8 @ f54bf39547 ›
Image: vllm/vllm-openai:v0.30.0@sha256:8a69ffad015f138d7170c4ddc429e230a3bc1c1719f67e14324749df200a4b90
Engine profile: vllm ›
Recipe file: registry/recipes/nvidia/rtx-pro-6000-blackwell-96gb/lfm2.5-24b-a2b.vllm.128k.json ›
Model released: Oct 12, 2026
Tested: owner, RTX PRO 6000 Blackwell WS (450 W cap)