Liquid AI
Category: Foundation Models / LLMs
Liquid AI develops efficient foundation models using liquid neural networks, enabling high-performance AI deployment on edge devices with minimal computational resources Liquid AI was founded in 2023. The company is led by Ramin Hasani. Based in Cambridge, United States. Team size: 51-200. Total funding raised: $287.6M. Latest round: Series A. Key investors include AMD (lead investor), Samsung Catalyst Fund, OSS Capital, PagsGroup, Duke Capital Partners.
- Founded
- 2023
- Headquarters
- Cambridge, United States
- Team size
- 51-200
- Total funding
- $287.6M
Value proposition
Delivers GPT-4o-class performance with 350M-2.6B parameter models that run entirely on-device, offering 3x training efficiency, zero cloud dependencies, and specialized task-specific optimization
Products and solutions
Liquid Foundation Models (LFMs), LFM2.5 on-device family (230M–8B-A1B MoE, Thinking, VL, Retrievers), LFM2-24B-A2B, Liquid Nanos / task-specific models, LEAP Edge AI Platform, Apollo Mobile App, Pipette on-device benchmarking suite, Custom / domain LFMs (e.g. LFM2-2.6B-MMAI)
Unique value
Pioneered liquid neural networks inspired by biological systems and dynamical systems theory, enabling real-time adaptive AI models that maintain high performance with minimal computational resources
Target customer
Enterprises requiring privacy-first, low-latency AI solutions across consumer electronics, automotive, healthcare, finance, and embedded systems
Industries served
Consumer Electronics, Automotive, Healthcare, Finance, Embedded Systems, Robotics, E-commerce
Technology advantage
Non-transformer architecture based on liquid neural networks offers superior efficiency, real-time adaptability, and 3x training efficiency improvement over traditional models while maintaining GPT-4o-class performance on specialized tasks
How they differentiate
Liquid AI differentiates through liquid neural networks architecture, extreme efficiency (350M-2.6B parameters vs competitors' 100B+), on-device deployment capability, real-time adaptability, and specialized task-specific models (Nanos)
Main competitors
OpenAI (GPT models), Anthropic (Claude), Google (Gemini), Microsoft (Copilot), Meta (Llama models)
Key partnerships
AMD (strategic partner and lead investor), Shopify (multi-year partnership for sub-20ms models), Mercedes-Benz (multi-year partnership for embedded in-car intelligence), Amazon Web Services (Bedrock/Marketplace), Hugging Face (model distribution), G42 (commercial partnership), Capgemini (enterprise collaboration), Insilico Medicine (strategic partnership for drug-discovery LFMs), MacPaw (long-term partnership for on-device Mac AI / Eney)
Notable customers
Shopify (multi-year partnership; 1B+ LFM requests by Jul 2026), AMD (strategic partnership), Mercedes-Benz (multi-year partnership for in-car AI), Insilico Medicine (drug-discovery LFMs), MacPaw (on-device Mac AI / Eney), Various enterprise clients in consumer electronics and automotive sectors
Major milestones
2023: Company founded by MIT CSAIL researchers, December 2024: Closed $250M Series A led by AMD at ~$2.3B valuation, July 2025: Released LFM2 on-device foundation models, September 2025: Launched Liquid Nanos with GPT-4o-class performance, November 2025: Announced multi-year partnership with Shopify, January 2026: Released LFM2.5 next-generation on-device models, February 2026: Launched LFM2-24B-A2B, March 2026: Strategic partnership with Insilico Medicine for drug-discovery LFMs, April 2026: Multi-year Mercedes-Benz partnership for embedded in-car intelligence, May 2026: Released LFM2.5-8B-A1B on-device MoE, July 2026: Shopify LFMs surpassed 1 billion platform requests, August 2026: MacPaw partnership for on-device Mac AI; shipped LFM2.5-2.6B agents, LFM2.5-VL-3B, Pipette
Growth metrics
Achieved $13.2M revenue (Nov 2025 est.); ~120–139 employees (PitchBook/LinkedIn mid-2026); LFM2.5 family + Pipette shipped; Shopify LFMs >1B requests (Jul 2026); MacPaw Mac on-device partnership (Aug 2026); Mercedes-Benz in-car AI production path H2 2026
Market positioning
Positioned as the leading edge-AI infrastructure company specializing in ultra-efficient, on-device foundation models for privacy-critical and low-latency applications
Geographic focus
Global with headquarters in Boston, Massachusetts, serving North American, European, and Asian markets through partnerships with AMD, Shopify, and cloud providers
Patents and IP
Proprietary liquid neural network architecture and training methodologies; specific patents not publicly disclosed
About Ramin Hasani
Co-founder and CEO of Liquid AI; previously Principal AI and Machine Learning Scientist at Vanguard Group and Research Affiliate at MIT CSAIL; PhD in Computer Science from Vienna University of Technology (TU Wien)
Latest news about Liquid AI
More Foundation Models / LLMs companies
Official website: https://www.liquid.ai
