**[3B Small Model Outperforms 8B: Liquid AI's New Vision Model Runs Locally]** According to monitoring by Beating, Liquid AI has released the vision-language model LFM2.5-VL-3B, featuring only 3.1 billion parameters and optimized for local devices such as smartphones and computers. It can understand web pages, app interfaces, documents, and multiple images, and has added the ability to invoke visual tools. Official testing across 28 visual benchmarks yielded an average score of 69.4. In comparison, the 8B-parameter Gemma-4-E4B scored only 59.7, while the 5.1B Gemma-4-E2B scored 52.0. Its performance is on par with the 4.7B-parameter InternVL 3.5 4B, and just 0.7 points lower than Qwen3.5-4B. Speed is its standout feature. After 4-bit quantization, the model can generate 228 tokens per second on an M5 Max, using approximately 3.3GB of memory. On the Galaxy S26 Ultra, it achieves 20 tokens per second. Under high concurrency on a single H100, the model's output throughput reaches up to 11,000 tokens per second, roughly double that of some 4B-level models. The model's weights have been made publicly available for download on Hugging Face and support llama.cpp, MLX, vLLM, SGLang, and ONNX. It is licensed under the LFM Open License v1.0, allowing free commercial use for annual revenue below $10 million, with additional commercial licensing required for higher revenue. [Original Link]