{"version":1,"type":"story","url":"https://digestai.news/story/liquid-ai-releases-lfm2-5-vl-dspark-for-faster-vision-language-model-i","json":"https://digestai.news/story/liquid-ai-releases-lfm2-5-vl-dspark-for-faster-vision-language-model-i.json","markdown":"https://digestai.news/story/liquid-ai-releases-lfm2-5-vl-dspark-for-faster-vision-language-model-i.md","slug":"liquid-ai-releases-lfm2-5-vl-dspark-for-faster-vision-language-model-i","headline":"Liquid AI releases LFM2.5-VL-DSpark for faster vision-language model inference","summary":"Liquid AI announced **LFM2.5-VL-DSpark**, a draft model for its **LFM2.5-VL-3B** vision-language model. The model uses speculative decoding to speed up inference by up to **3.13x on-device** and **2.66x on H100 GPUs**, with end-to-end gains of **2.62x and 2.27x**, respectively. It adds **280M parameters** (8.9% increase) but maintains output quality, according to the company’s benchmarks across tasks like VQA, captioning, and multi-turn conversation.\n\nThe model supports **day-one integration** with **llama.cpp, MLX-VLM, and SGLang**, targeting edge and GPU deployments. Liquid AI emphasizes its open-weight approach, allowing unrestricted fine-tuning and deployment. The release aligns with the lab’s goal of AI running across devices, from base models to specialized variants like audio and vision.","keyPoints":["LFM2.5-VL-DSpark speeds up vision-language model inference by up to 3.13x on-device and 2.66x on H100 GPUs","Adds 280M parameters (8.9% increase) to LFM2.5-VL-3B with no output quality trade-off, per Liquid AI","Supports day-one integration with llama.cpp, MLX-VLM, and SGLang for edge and GPU deployment"],"whyItMatters":"Faster vision-language model inference reduces latency for applications like real-time captioning and multi-modal reasoning, critical for edge devices and latency-sensitive workflows.","category":{"slug":"models","name":"Generative AI & Models","url":"https://digestai.news/category/models"},"entities":{"companies":["Liquid AI","Hugging Face"],"models":["LFM2.5-VL-3B","LFM2.5-VL-DSpark"],"people":[]},"firstPublishedAt":"2026-09-24T14:08:57Z","updatedAt":"2026-09-24T14:08:57Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"Hugging Face","title":"Accelerating vision-language models with LFM2.5-VL-DSpark","url":"https://huggingface.co/blog/LiquidAI/lfm2-5-vl-dspark","publishedAt":"2026-09-24T14:08:57Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":{"title":"Rise of Local Open-Source AI Tools","url":"https://digestai.news/thread/kdnuggets-lists-seven-open-source-chatgpt-alternatives-that-run-locally","storyCount":3},"cite":{"text":"Digest AI, \"Liquid AI releases LFM2.5-VL-DSpark for faster vision-language model inference\", 24 September 2026, https://digestai.news/story/liquid-ai-releases-lfm2-5-vl-dspark-for-faster-vision-language-model-i","publisher":"Digest AI","title":"Liquid AI releases LFM2.5-VL-DSpark for faster vision-language model inference","datePublished":"2026-09-24T14:08:57Z","url":"https://digestai.news/story/liquid-ai-releases-lfm2-5-vl-dspark-for-faster-vision-language-model-i"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}