LFM2.5

Liquid AI🇺🇸 USMajor lab· Jun 3, 2026

TextReasoningopen weightsAPI

LFM2.5 is a hybrid model designed for on-device deployment, featuring 230 million parameters and optimized for fast edge inference. It is suitable for general-purpose text generation and agentic tasks, but not recommended for reasoning-heavy workloads.

Summary of the model card

Context: 66K tokens
Price / 1M tok: free in · free out
License: other
Modalities: text → text

34 variants

2.6B Free	2.6B	66K	free/free	OR
8B A1B Dspark GGUF	8B-A1B			HF
2.6B Dspark GGUF	2.6B			HF
1.2B Instruct Dspark GGUF	1.2B			HF
2.6B Dspark	2.6B			HF
8B A1B Dspark	8B-A1B			HF
1.2B Instruct Dspark	1.2B			HF
2.6B MLX NVFP4	2.6B			HF
2.6B MLX MXFP4	2.6B			HF
2.6B MLX MXFP8	2.6B			HF
2.6B MLX 4BIT	2.6B			HF
2.6B MLX 5BIT	2.6B			HF
2.6B MLX 6BIT	2.6B			HF
2.6B MLX 8BIT	2.6B			HF
2.6B MLX BF16	2.6B			HF
2.6B MLX	2.6B			HF
2.6B Onnx	2.6B			HF
2.6B GGUF	2.6B			HF
2.6B Base	2.6B			HF
2.6B	2.6B			HF
230M	230M			HF
230M Base	230M			HF
230M Onnx	230M			HF
230M MLX 4BIT	230M			HF
230M MLX 5BIT	230M			HF
230M MLX 6BIT	230M			HF
230M MLX 8BIT	230M			HF
230M MLX BF16	230M			HF
230M GGUF	230M			HF
1.2B Jp 202606 Onnx	1.2B			HF
1.2B Jp 202606 MLX 5BIT	1.2B			HF
1.2B Jp 202606 MLX 4BIT	1.2B			HF
1.2B Jp 202606 MLX 6BIT	1.2B			HF
1.2B Jp 202606 MLX 8BIT	1.2B			HF

Sources

HuggingFace OpenRouter Paper

Deutsch

LFM2.5 ist ein hybrides Modell, das für die On-Device-Bereitstellung konzipiert ist, mit 230 Millionen Parametern und optimiert für schnelle Edge-Inferenz. Es eignet sich für allgemeine Textgenerierung und agentische Aufgaben, wird jedoch nicht für reasoning-intensive Arbeitslasten empfohlen.

Provenance

Detected: Sep 27, 2026
First source: huggingface
Description: Summary of the model card
Editorially reviewed: —