# Liquid AI

> Liquid AI is an MIT CSAIL spinout building Liquid Foundation Models (LFMs), a family of general-purpose AI models engineered to run fast and privately on-device - phones, laptops, wearables, and cars - instead of relying on the cloud. Rooted in the founders' 'liquid neural network' research inspired by the nervous system of a roundworm, the company pairs a first-principles model architecture with LEAP, a developer platform that deploys models into mobile apps in a few lines of code, and Apollo, a fully local AI chat app. It raised a $250M Series A led by AMD in December 2024 at a roughly $2.35B valuation.

- **Founded:** 2023
- **Headquarters:** Cambridge, Massachusetts, United States
- **Founders:** Ramin Hasani (Co-founder & CEO), Mathias Lechner (Co-founder & CTO), Alexander Amini (Co-founder & Chief Scientific Officer), Daniela Rus (Co-founder & Board Member; Director of MIT CSAIL)
- **Team size:** ~100 employees
- **Products:** Liquid Foundation Models (LFMs), LFM2 / LFM2.5, LEAP (Liquid Edge AI Platform), Liquid Apollo, Liquid Playground / Hugging Face models
- **Notable:** Spun out of MIT CSAIL on the back of the 'Liquid Time-Constant Networks' research line., Raised a $250M Series A led by AMD at a ~$2.35B valuation in December 2024., Released LFM2/LFM2.5, positioned as among the fastest on-device foundation models on the market.

## Products & services

- **Liquid Foundation Models (LFMs)** — A family of general-purpose generative models built on a first-principles architecture (rooted in liquid neural networks) rather than a standard transformer, designed for strong quality at low memory and compute cost.
- **LFM2 / LFM2.5** — Second-generation on-device foundation models optimized for speed and memory efficiency, spanning base, instruct, vision-language, and audio-language variants for real on-device agents.
- **LEAP (Liquid Edge AI Platform)** — A developer platform and SDK to deploy foundation models into iOS and Android apps in roughly 10 lines of code, without cloud infrastructure.
- **Liquid Apollo** — A fully local, privacy-preserving AI chat mobile app that runs LEAP's model library on-device with no data leaving the phone.
- **Liquid Playground / Hugging Face models** — Open-weight models and a hosted playground for testing and evaluating LFMs.

## Achievements

- Spun out of MIT CSAIL on the back of the 'Liquid Time-Constant Networks' research line.
- Raised a $250M Series A led by AMD at a ~$2.35B valuation in December 2024.
- Released LFM2/LFM2.5, positioned as among the fastest on-device foundation models on the market.
- Launched LEAP and Apollo to make on-device AI deployable in ~10 lines of code.
- Reported production edge deployments with partners including Mercedes-Benz.

## Latest updates

- **2026-01** — Released LFM2.5, a compact model family (Base, Instruct, Japanese, Vision-Language, Audio-Language) built for real on-device agents, with pretraining scaled toward 28T tokens and a reinforcement-learning post-training pipeline.
- **2025-07** — Launched LEAP and Liquid Apollo, bringing on-device AI deployment and a fully local chat app to developers.
- **2025** — Introduced LFM2 and multimodal variants including LFM2-VL vision-language models for the edge.
- **2024-12** — Closed a $250M Series A led by AMD at a ~$2.35B valuation.

## Links

- Website: https://www.liquid.ai
- LinkedIn: https://www.linkedin.com/company/liquid-ai-inc
- Twitter/X: https://twitter.com/LiquidAI_
- GitHub: https://github.com/Liquid4All
- YouTube: https://www.youtube.com/@LiquidAI_

---

Profile page: https://yespress.io/liquid-ai
Published by YesPress — https://yespress.io
Last updated: 2026-07-19