LLM Reference

LFM-2 Models by Liquid AI

Liquid AILFM Open License v1.0Open weights
2 models2025Up to 32k ctxFrom $0.03/1M input

Last refreshed 2026-07-10. Next refresh: weekly.

Details

ResearcherLiquid AI
Commercial useCommercial use: conditional
Models2
Released2025
Max context32k

Capabilities

JSON / Tool use1 of 2 models
Structured Outputs1 of 2 models

Links

Website

About

LFM-2 (Liquid Foundation Model 2) is Liquid AI's second-generation hybrid architecture model series, combining attention-based and structured state-space layers for efficient long-context inference.

Decision facts

Best fit
JSON / Tool usestructured outputs
Capability starting point
LFM2-24B-A2B with 32k context and JSON / Tool use and structured outputs
Lowest tracked input
LFM2-24B-A2B · $0.03/1M · OpenRouter
Closest related family
LFM-2.5

Current Variants

Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.

2 in view

Use when the workload needs 8k context and 23.8B parameters.

2025-128k context23.8B parameters

Use when the workload needs 32k context, 24B parameters, and JSON / Tool use.

2025-1132k context24B parametersJSON / Tool use

Release Timeline

2 release groups
2025-12
1 current
Together LFM2-24B
8k context23.8B parameters
Current
2025-11
1 current
LFM2-24B-A2B
32k context24B parametersJSON / Tool use
Current

Specifications(2 models)

LFM-2 model specifications comparison
ModelReleasedContextParametersJSON / Tool useStructured Outputs
Together LFM2-24B2025-128k23.8BNoNo
LFM2-24B-A2B2025-1132k24BYesYes

Available From(1 provider)

Pricing

LFM-2 model pricing by provider
ModelProviderInput / 1MOutput / 1MType
LFM2-24B-A2BOpenRouter$0.03$0.12Serverless