{"slug": "liquid-ai-releases-six-lfm3-models-for-on-device-specialization", "title": "Liquid AI releases six LFM3 models for on-device specialization", "summary": "Liquid AI published its LFM3 announcement on Hugging Face on October 27th, releasing six models spanning text, vision, audio, omni and decision tasks, all built from a shared base and designed for further specialization. Liquid AI says each model can be fine-tuned on a single GPU, and the checkpoints are available on Hugging Face. The release puts co-founder and CEO Ramin Hasani's device-first strategy into a developer-facing family, though adoption will depend on performance in real workloads rather than the company's benchmark claims.", "body_md": "# Liquid AI releases six LFM3 models for on-device specialization\n\n**The shared-base family covers text, vision, audio and decision tasks. Liquid AI says each model can be fine-tuned on a single GPU, and its checkpoints are available on Hugging Face.**\n\n        By [Ryan Merket](https://runtimewire.com/author/ryan-merket)\n        · Published \n\nPrimary source: [Hugging Face Newsroom](https://huggingface.co/blog/LiquidAI/lfm3)\n\n## Why it matters\n\nLFM3 puts Liquid AI's device-first strategy into a developer-facing model family: shared foundations, task-specific fine-tuning and local deployment. Whether that translates into customer adoption depends on performance in real workloads, not the company's benchmark claims alone.\n\n[Ramin Hasani](https://www.liquid.ai/team/ramin-hasani?ref=runtimewire) ([@ramin_m_h](https://x.com/ramin_m_h?ref=runtimewire)), the co-founder and CEO of [Liquid AI](https://www.liquid.ai/?ref=runtimewire), has spent years making the case that useful AI should not have to live in a data center. On October 27th, Liquid AI published [its LFM3 announcement on Hugging Face](https://huggingface.co/blog/LiquidAI/lfm3?ref=runtimewire): six text, vision, audio, omni and decision models built from a shared base and designed for further specialization. The announcement was published that day.…\n\nWhen you purchase through links in our articles, [we may earn a small commission](https://runtimewire.com/affiliate-monetization-standards). This doesn’t affect our editorial independence.", "url": "https://wpnews.pro/news/liquid-ai-releases-six-lfm3-models-for-on-device-specialization", "canonical_source": "https://runtimewire.com/article/liquid-ai-lfm3-specialized-multimodal-models", "published_at": "2026-10-27 23:37:19+00:00", "updated_at": "2026-10-28 02:42:37.390135+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-products", "ai-tools", "ai-infrastructure"], "entities": ["Liquid AI", "Ramin Hasani", "LFM3", "Hugging Face"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/liquid-ai-releases-six-lfm3-models-for-on-device-specialization", "markdown": "https://wpnews.pro/news/liquid-ai-releases-six-lfm3-models-for-on-device-specialization.md", "text": "https://wpnews.pro/news/liquid-ai-releases-six-lfm3-models-for-on-device-specialization.txt", "jsonld": "https://wpnews.pro/news/liquid-ai-releases-six-lfm3-models-for-on-device-specialization.jsonld"}}