LFM2 24B A2B

Name: LFM2 24B A2B
Author: Liquid AI

by Liquid AI

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released November 2025128K context≈ 96,000 words24B params

LiquidAI's hybrid architecture model that blends transformer and liquid neural network components, making it unusually efficient for its parameter count — 24B total but only 2B active at inference time. It handles long contexts up to 128K tokens while keeping compute costs low, a trade-off that shows in strong coherence over extended documents.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Creative Writing

Moderate

Factual Knowledge

LFM2 24B A2B

by Liquid AI

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released November 2025128K context≈ 96,000 words24B params

LiquidAI's hybrid architecture model that blends transformer and liquid neural network components, making it unusually efficient for its parameter count — 24B total but only 2B active at inference time. It handles long contexts up to 128K tokens while keeping compute costs low, a trade-off that shows in strong coherence over extended documents.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Creative Writing

Moderate

Factual Knowledge

Glossary

ArchitectureThe underlying structural design of a neural network that defines how data flows through layers and components.CoherenceThe quality of maintaining consistent meaning and logical flow across multiple sentences or exchanges in a conversation.Hybrid ArchitectureA model that combines two different neural network designs (in this case, Mamba2 and attention mechanisms) to balance speed and performance.InferenceThe process of running a trained model to generate predictions or outputs from new inputs.Inference TimeThe amount of time it takes for a model to process input and generate output after it has been trained.Liquid Neural NetworkA neural network architecture that uses continuous, adaptive functions instead of fixed layers, allowing the model to dynamically adjust its computation based on input.Parameter CountThe total number of adjustable weights in a model; more parameters generally mean more capacity to learn, but also require more computing power.TokensThe basic units of text that a language model processes, typically representing words or word fragments.TransformerThe dominant neural network architecture for language models, using self-attention to process sequences.

LFM2 24B A2B

Capabilities

LFM2 24B A2B

Capabilities

Use Case Fit

Similar Models

Glossary