LLM Reference

LFM2.5 8B A1B

Released
2026-05-28
Last refreshed
2026-07-10
Status
Researched 3d ago
Open weightsCommercial use: conditionalRAGAgentsLong contextClassificationJSON / Tool use

LFM2.5 8B A1B is a released rag, agents, and long context model with open-weight and 128k context; evaluate it while provider pricing coverage matures.

Use it for

  • Teams evaluating rag, agents, and long context
  • Workloads that can use a 128k context window

Do not use it for

  • Cost-sensitive launches that need sourced token pricing
  • Vision or document-understanding workloads
  • Teams that need a tracked hosted API route today
Specifications
Family
LFM-2.5
Released
2026-05-28
Context
128k
Max output
8,192
Parameters
8.3B
Architecture
Mixture of Experts
Specialization
general
Openness
Open weights
License
LFM Open License v1.0Commercial use: conditional
Weights
Available
Code
Unknown
Training
Pretrained
Created by

Adaptive and efficient AI models

Cambridge, MA, United States
Founded 2023
Website
Pricing

No tracked provider token pricing is available yet.

About

LFM2.5-8B-A1B is Liquid AI's latest on-device mixture-of-experts model, succeeding LFM2-8B-A1B. It has 8.3B total parameters with approximately 1.5B active per token (the A1B label uses a rounded ~1B figure). The architecture combines 18 double-gated LIV convolutional layers with 6 GQA attention layers, trained on 38 trillion tokens. The context window expands to 128K tokens (up from 32K in the predecessor). It is a reasoning model that generates explicit chain-of-thought steps before producing its final answer, making reasoning tokens cheap due to the MoE design. Strong tool-calling, function-calling, and instruction-following capabilities make it well-suited for agentic workflows on edge hardware. Weights are openly available on Hugging Face under the lfm1.0 license.

LFM2.5 8B A1B is an open-weight model in the LFM-2.5 family. The structured metadata tracks a 128k-token context window, reasoning, function calling, tool use, and structured outputs. Headline tracked benchmarks include Instruction-Following Evaluation 91.8, MATH-500 88.8, and AIME 2025 42.5.

Top use-case fit: coding, agents, and build tasks

RAG

Included by capability and metadata signals in the decision map.

Agents

1 relevant benchmark in the decision map.

Long context

Included by capability and metadata signals in the decision map.

Provider price ladder

No tracked provider token pricing is available for this model yet.

Capabilities

ReasoningFunction CallingTool UseStructured Outputs

Benchmark peer barsfor Agents

Benchmark scores(6)

Scores are benchmark-specific and are direction-aware: the same numeric gap can mean very different outcomes across suites. Use the leaderboard context and this model's provider route to decide whether the winning margin is meaningful for your workload.
BenchmarkScoreVersionSource
Instruction-Following Evaluation91.8https://www.liquid.ai/blog/lfm2-5-8b-a1b
MATH-50088.8MATH-500 (accuracy)https://www.marktechpost.com/2026/05/28/liquid-ai-releases-lfm2-5-8b-a1b-an-on-device-moe-model-with-8-3b-total-and-1-5b-active-parameters/
AIME 202542.5https://www.liquid.ai/blog/lfm2-5-8b-a1b
Google-Proof Q&A34.4GPQA Diamond (accuracy)https://benchlm.ai/models/lfm2-5-8b-a1b
MMLU PRO50.5Third-party evaluation (accuracy)https://benchlm.ai/models/lfm2-5-8b-a1b
τ-bench88.1τ² Telecom benchmark (accuracy)https://www.marktechpost.com/2026/05/28/liquid-ai-releases-lfm2-5-8b-a1b-an-on-device-moe-model-with-8-3b-total-and-1-5b-active-parameters/

Migration checks

No linked migration route is available for this model yet.

Frequently asked questions

What is the context window of LFM2.5 8B A1B?

LFM2.5 8B A1B has a context window of 128k tokens.

What is the max output of LFM2.5 8B A1B?

LFM2.5 8B A1B can generate up to 8,192 output tokens.

When was LFM2.5 8B A1B released?

LFM2.5 8B A1B was released on 2026-05-28.

What benchmarks has LFM2.5 8B A1B been tested on?

LFM2.5 8B A1B has been evaluated on 6 benchmarks, including Instruction-Following Evaluation, MATH-500, AIME 2025, Google-Proof Q&A, MMLU PRO.