Models / Liquid AI
Chat
Reasoning

LFM2.5-8B-A1B

Compact MoE model with fast tool calling and a sharpened knowledge boundary

Try now
read docs

About model

LFM2.5-8B-A1B is Liquid AI's compact Mixture-of-Experts model, built on the LFM2 architecture combining MoE routing, grouped-query attention, and gated short convolution blocks. It extends the prior LFM2-8B-A1B release with a 128K-token context window, a doubled 128K vocabulary for more efficient tokenization of non-Latin scripts, and large-scale reinforcement learning, and it now produces an explicit reasoning trace before its final answer. LFM2.5-8B-A1B reaches 91.84% on IFEval, 88.76% on MATH-500, and 88.07% on Tau2-Bench Telecom, competitive with much larger dense and MoE models on instruction following and agentic tool-use benchmarks. A targeted reinforcement learning stage sharpens its knowledge boundary, reducing hallucination while preserving accuracy. Available on Together AI.

‍

IFEval

91.84%

Strong instruction following, competitive with much larger MoE models

Tau2-Bench Telecom

88.07%

Reliable multi-step tool calling in agentic workflows

Context Window

128K

Expanded from 32K for longer documents and reasoning traces

Model key capabilities
  • Fast Tool Calling: 88.07% Tau2-Bench Telecom and 64.36% BFCLv3, built for chaining tool calls in agentic workflows
  • Strong Instruction Following: 91.84% IFEval and 56.47% IFBench, competitive with much larger dense and MoE models
  • Sharpened Knowledge Boundary: Reinforcement learning stage that reinforces abstention on unreliable queries, reducing hallucination
  • Production-Ready Infrastructure: 99.9% SLA, available on serverless and dedicated infrastructure
Performance benchmarks

Model

AIME 2025

GPQA Diamond

HLE

LiveCodeBench

MATH500

SWE-bench verified

LFM2.5-8B-A1B

42.53

88.76

Related open-source models

Competitor closed-source models

Claude Opus 4.6

90.5%

34.2%

78.7%

OpenAI o3

83.3%

24.9%

99.2%

62.3%

OpenAI o1

76.8%

96.4%

48.9%

GPT-4o

49.2%

2.7%

32.3%

89.3%

31.0%

  • API usage

    • cURL
    • Python
    • Typescript

    Endpoint:

    LiquidAI/LFM2.5-8B-A1B

  • Model card

    Architecture Overview:
    • LFM2 architecture combining Mixture-of-Experts routing, grouped-query attention, and gated short convolution blocks
    • 8.5B total parameters with 1B active per token
    • 128K-token context window, expanded from 32K in the prior LFM2-8B-A1B release
    • 128K vocabulary, doubled from 65K, improving tokenization efficiency for Hindi, Thai, Vietnamese, Indonesian, and Arabic
    • Reasoning-only model that produces an explicit chain of thought before its final answer

    Training Methodology:
    • Extended pre-training scaled from 12T to 38T tokens versus the prior LFM2-8B-A1B release, followed by large-scale reinforcement learning
    • Context extended in stages: a 2T-token midtraining phase to 32K focused on reasoning, math, and tool use, then a 400B-token stage to 128K focused on long-document and long-trajectory data
    • A targeted preference optimization stage reduces looping behavior in long reasoning traces
    • A reinforcement learning stage using an avg@k-based reward over a diverse knowledge dataset reinforces abstention on queries beyond reliable knowledge

    Performance Characteristics:
    • Knowledge and instruction following: -24.70 AA-Omniscience Index, 91.84% IFEval, 56.47% IFBench, 79.93% Multi-IF
    • Math: 88.76% MATH-500, 42.53% AIME 2025, 50.00% AIME 2026
    • Tool use and agentic: 64.36% BFCLv3, 48.50% BFCLv4, 88.07% Tau2-Bench Telecom, 39.82% Tau2-Bench Retail
    • Liquid AI reports the model matches larger MoE models such as Gemma 4-26B-A4B on instruction-following benchmarks at a fraction of the active parameter count

    ‍

  • Prompting

    Together AI API Access:
    • Access LFM2.5-8B-A1B via Together AI APIs using the endpoint LiquidAI/LFM2.5-8B-A1B
    • Authenticate using your Together AI API key in request headers
    • Supports JSON mode and tool calling for structured, multi-step agentic workflows
    • The model produces an explicit reasoning trace before its final response
    • Available on Together AI serverless and dedicated infrastructure

    ‍

  • Applications & use cases

    Agentic Tool-Calling Workflows:
    • Chain multi-step tool calls for tasks that require sequenced actions and confirmations
    • Build telecom and retail-style agent flows that lean on the model's Tau2-Bench strength
    • Combine tool calling with JSON mode for structured downstream integrations

    Instruction-Heavy Assistants:
    • Follow complex, multi-part instructions reliably across conversation turns
    • Run assistant workflows that require precise formatting and constraint adherence
    • Handle multilingual instructions with tokenization efficient across non-Latin scripts

    Cost-Efficient Reasoning at Scale:
    • Serve reasoning-backed responses from a compact active parameter count for high-volume workloads
    • Run math and structured problem-solving tasks through the Together endpoint
    • Scale concurrent agent sessions with the model's compact active footprint

    ‍

Model specifications
Model data
  • Model provider
    Liquid AI
  • Features
    JSON Mode
  • Deployment
    Serverless
    On-Demand Dedicated
  • Endpoint
    LiquidAI/LFM2.5-8B-A1B
  • Parameters
    8.5B
  • Activated parameters
    1B
  • Context length
    128K
  • Input price

    $0.03 / 1M tokens

  • Output price

    $0.12 / 1M tokens

  • Input modalities
    Text
  • Output modalities
    Text
  • Released
    May 27, 2026
Run in Playground
Quickstart docs
Deploy model