Liquid AI
Liquid AI
Language model · United States ·May 2026

LFM2.5-8B-A1B

Language Open Weights
8B
Parameters
128K
Context
3
Benchmarks

About LFM2.5-8B-A1B

LFM2.5-8B-A1B is an AI model developed by Liquid AI, in the language category, released in 2026, made available as an open-weights model with 8B params. An edge-optimized sparse Mixture-of-Experts model with 8B total parameters and about 1B active per token, built for fast.

On this page you'll find LFM2.5-8B-A1B's full specifications, including a context window of 128K tokens, 3 benchmark results. Review provider pricing and benchmark scores below, or compare LFM2.5-8B-A1B head-to-head with other language models.

→ More models by Liquid AI → All language models → AI benchmarks leaderboard

Run LFM2.5-8B-A1B locally — what to buy

Share this build ↗

Sized to what LFM2.5-8B-A1B actually needs (~128k context at Q4), not to the biggest GPU: Good = cheapest that runs it, Better = best value, Best = most headroom. Speed is a hardware estimate; anything past ~120 tok/s is shown as instant.

Good
Intel Arc B580 12GB
Consumer GPU · ~$249
~73.3 tok/s
Q4 · fits full context · up to 39k ctx
Runs with: ExLlamaV2, KoboldCpp, LM Studio
See full estimate →
Better
Intel Arc A770 16GB
Consumer GPU · ~$329
~90 tok/s
Q4 · fits full context · up to 67k ctx
Runs with: ExLlamaV2, KoboldCpp, LM Studio
See full estimate →
Best
NVIDIA GeForce RTX 5070
Consumer GPU · ~$549
~120 tok/s
Q4 · fits full context · up to 39k ctx
Runs with: ExLlamaV2, KoboldCpp, LM Studio
See full estimate →

Software support

·ExLlamaV2 ·KoboldCpp ·LM Studio ·MLX ·Ollama ·SGLang ·TensorRT-LLM ·llama.cpp ·vLLM ?KTransformers ?text-generation-webui

✓ measured · · compatible · — not supported. Informational only — speed is hardware-based.

Consumer GPU
36–120+ tok/s
32 options →
Apple
21.8–120+ tok/s
18 options →
Datacenter GPU
120+ tok/s
13 options →

Community

Would you run LFM2.5-8B-A1B again?
No votes yet — be the first.
Did it run? — community reports by hardware

No reports yet. Own this model on some hardware? Be the first to confirm it runs.

Discussion

No comments yet. Start the discussion.

LFM2.5-8B-A1B benchmark scores▶
Compare on the benchmark table →
Benchmark Category Score Variant
IFEval general 91.84%
MATH-500 math 88.76%
BFCL v3 tool-use 64.36%

More models like LFM2.5-8B-A1B

AI
DMind-1
 AI
COLLECTIVE-1
 AI
Cogito 1 series
 AI
Euler (Euler大模型)
· 2.5B
 AI
SmolLM-1.7B
· 1.7B
 AI
Llama-3-70B-Synthia-v3.5
· 70B

Frequently asked questions

What is LFM2.5-8B-A1B? ▶

An edge-optimized sparse Mixture-of-Experts model with 8B total parameters and about 1B active per token, built for fast It is tracked on GenAIList with its specifications, benchmark scores and provider pricing.

Who created LFM2.5-8B-A1B? ▶

LFM2.5-8B-A1B was developed by Liquid AI and released in 2026.

Is LFM2.5-8B-A1B open source or proprietary? ▶

LFM2.5-8B-A1B ships as an open-weights model: you can download and self-host the weights, though the licence may place some restrictions on use.

What is LFM2.5-8B-A1B's context window? ▶

LFM2.5-8B-A1B supports a context window of 128K tokens, and has 8B params.

How much does LFM2.5-8B-A1B cost? ▶

Pricing for LFM2.5-8B-A1B depends on the provider. See the providers table on this page for the latest API rates.

How does LFM2.5-8B-A1B perform on benchmarks? ▶

LFM2.5-8B-A1B is benchmarked across 3 evaluations on GenAIList, including IFEval (91.84%). See the full benchmark table below and compare it with other models.

LFM2.5-8B-A1B — Fastest Local AI Agent on a Laptop? (6 Tests)
Releasing LFM2.5-8B-A1B, a model for local agents
Liquid AI's new model LFM2.5-8B-A1B debuts: 1.5B active
[ES] Liquid LFM2.5-8B-A1B: el modelo de 8B que rivaliza con modelos 4x mas grandes

Know a great video about LFM2.5-8B-A1B?