LFM3-VL-2B

Use it to analyze images and video, answer visual questions, and support applications that require spatial reasoning.

Availability

The base and post-trained models are available today on Hugging Face, Liquid AI Playground, and Liquid AI API

Pricing model

Not established

Deployment

Designed for efficient deployment; models can be fine-tuned locally

Privacy

Not established

Licence

Open weights

• The LFM3 base and post-trained models are available today on Hugging Face, Liquid AI Playground, and Liquid AI API.
• LFM3-VL-2B takes video as input. ^{[1] [2] [3]}

  Video understanding: for the first time, an LFM vision model takes video as input.

• The model reasons step by step before answering. ^{[1] [2] [3]}

  Thinking: the model reasons step by step before answering.

• The model can localize objects and reason about spatial relationships. ^{[1] [2] [3]}

  Spatial and embodied reasoning: the model’s ability to localize objects and reason about their spatial relationships provides a strong foundation for downstream robotics applications.

• The source reports that LFM3-VL-2B scored significantly ahead of similarly sized models across 62 vision benchmarks. ^[1]

  Across 62 vision benchmarks, LFM3-VL-2B scores significantly ahead of all models of similar size by at least 7.4 points, and despite its size, it matches the performance of Gemma 4 12B, a model more than 6 times larger, surpasses Gemini 3.5 Flash Lite and is within 2 points of GPT-5.6 Luna.