# Liquid AI

> Liquid AI is an efficiency-first foundation model company spun out of MIT CSAIL in 2023, building Liquid Foundation Models (LFMs) and decision models designed to run efficiently on-device and at the edge rather than in cloud data centers.

Company page on Leadbay: https://mad-max.leadbay.ai/company/liquid.ai · loaded 2026-11-03

## At a glance

- Website: https://liquid.ai
- Team based in: Cambridge (liquid.ai and 1 more)
- Founded: 2023 (liquid.ai and 1 more)
- Founders: Ramin Hasani, Mathias Lechner
- Industry: Information Services
- Employees: 140 (employees and founders, investors not counted; liquid.ai and 1 more)
- Total funding: $293.2M (liquid.ai and 1 more)
- Backed by: AMD, OSS Capital, PagsGroup / Stephen Pagliuca, Duke Capital Partners, Breyer Capital, Tom Preston Werner / GitHub Co-founder, Naval Ravikant, Safar Partners, ISAI Cap Venture / Capgemini VC Fund, Tobias Lütke / Shopify Co-founder, David Siegel / Two Sigma, Chris Prucha / Notion Co-founder, Samsung Next, David Blundin / Link Ventures, Bold Capital Partners / Peter Diamandis, Marc Casper / Thermo Fisher, Argean Capital Management, Latham & Watkins, Bob Young / Red Hat Co-founder, Automattic
- Seen in its site's code: Next.js (/_next/static/), Google Workspace (MX smtp.google.com), HubSpot (TXT include:44218040.spf05.hubspot), Zoom (TXT ZOOM_verify) (read 2026-11-03)

## What it sells, to whom

- Liquid AI is an efficiency-first foundation model company spun out of MIT CSAIL in 2023, building Liquid Foundation Models (LFMs) and decision models designed to run efficiently on-device and at the edge rather than in cloud data centers. ([liquid.ai](https://www.liquid.ai/company), [liquid.ai](https://www.liquid.ai/), [liquid.ai](https://www.liquid.ai/blog/new-generation-of-ai-models-from-first-principles))

## Why it stands out

- d1-ultra: Top-scoring decision model on Artificial Analysis and Vals AI ClaimBench leaderboards (October 2026) ([liquid.ai](https://www.liquid.ai/api), [linkedin.com](https://www.linkedin.com/posts/raminhasani_from-the-localllama-community-on-reddit-activity-7428665260959686656-lRE7), [linkedin.com](https://www.linkedin.com/posts/daniela-rus-220b3_time100-ai-2026-daniela-rus-activity-7499092617884774400-IYM5))
- LFM2.5-1.2B: #1 in tool-calling judgement study by LocalLLaMA community (February 2026) ([liquid.ai](https://www.liquid.ai/api), [linkedin.com](https://www.linkedin.com/posts/raminhasani_from-the-localllama-community-on-reddit-activity-7428665260959686656-lRE7), [linkedin.com](https://www.linkedin.com/posts/daniela-rus-220b3_time100-ai-2026-daniela-rus-activity-7499092617884774400-IYM5))
- Daniela Rus: TIME100 AI 2026 honoree (August 2026) ([liquid.ai](https://www.liquid.ai/api), [linkedin.com](https://www.linkedin.com/posts/raminhasani_from-the-localllama-community-on-reddit-activity-7428665260959686656-lRE7), [linkedin.com](https://www.linkedin.com/posts/daniela-rus-220b3_time100-ai-2026-daniela-rus-activity-7499092617884774400-IYM5))
- Daniela Rus: 2026 High-Tech Prize of the Bavarian Minister-President (July 2026) ([liquid.ai](https://www.liquid.ai/api), [linkedin.com](https://www.linkedin.com/posts/raminhasani_from-the-localllama-community-on-reddit-activity-7428665260959686656-lRE7), [linkedin.com](https://www.linkedin.com/posts/daniela-rus-220b3_time100-ai-2026-daniela-rus-activity-7499092617884774400-IYM5))
- Daniela Rus: 2025 IEEE Edison Medal ([liquid.ai](https://www.liquid.ai/api), [linkedin.com](https://www.linkedin.com/posts/raminhasani_from-the-localllama-community-on-reddit-activity-7428665260959686656-lRE7), [linkedin.com](https://www.linkedin.com/posts/daniela-rus-220b3_time100-ai-2026-daniela-rus-activity-7499092617884774400-IYM5))
- Daniela Rus: 2024 John Scott Award ([liquid.ai](https://www.liquid.ai/api), [linkedin.com](https://www.linkedin.com/posts/raminhasani_from-the-localllama-community-on-reddit-activity-7428665260959686656-lRE7), [linkedin.com](https://www.linkedin.com/posts/daniela-rus-220b3_time100-ai-2026-daniela-rus-activity-7499092617884774400-IYM5))
- Competes with: Google (Gemini Nano on-device models), Microsoft (Phi small models), Meta (small Llama variants), Mistral AI (compact models), Qualcomm (on-device AI stack), Apple (on-device AI), OpenAI (GPT models), Anthropic (Claude models), Alibaba (Qwen models) ([yespress.io](https://yespress.io/liquid-ai), [neuronfeed.com](https://neuronfeed.com/startups/liquid-ai), [7wdata.be](https://7wdata.be/company/liquid/))

## Products and services

- Liquid Foundation Models (LFMs) - text, vision-language, audio, and omni-modal models (LFM2, LFM2.5, LFM3 families) ([liquid.ai](https://www.liquid.ai/models), [liquid.ai](https://www.liquid.ai/api), [yespress.io](https://yespress.io/liquid-ai))
- d1 decision models (d1, d1-ultra, d1-omni) for classification, routing, and scoring with calibrated probabilities ([liquid.ai](https://www.liquid.ai/models), [liquid.ai](https://www.liquid.ai/api), [yespress.io](https://yespress.io/liquid-ai))
- LEAP (Liquid Edge AI Platform) - developer SDK for deploying foundation models into iOS/Android apps ([liquid.ai](https://www.liquid.ai/models), [liquid.ai](https://www.liquid.ai/api), [yespress.io](https://yespress.io/liquid-ai))
- Liquid Apollo - fully local, privacy-preserving AI chat app ([liquid.ai](https://www.liquid.ai/models), [liquid.ai](https://www.liquid.ai/api), [yespress.io](https://yespress.io/liquid-ai))
- Liquid API - OpenAI-compatible API endpoint for LFM and decision models ([liquid.ai](https://www.liquid.ai/models), [liquid.ai](https://www.liquid.ai/api), [yespress.io](https://yespress.io/liquid-ai))
- Liquid Console - developer console for model access and fine-tuning ([liquid.ai](https://www.liquid.ai/models), [liquid.ai](https://www.liquid.ai/api), [yespress.io](https://yespress.io/liquid-ai))
- LFM2.5-Encoder models (230M, 350M) for retrieval, PII detection, and classification ([liquid.ai](https://www.liquid.ai/models), [liquid.ai](https://www.liquid.ai/api), [yespress.io](https://yespress.io/liquid-ai))
- Local AI builds (GGUF, MLX, ONNX) for laptops, phones, and edge boxes ([liquid.ai](https://www.liquid.ai/models), [liquid.ai](https://www.liquid.ai/api), [yespress.io](https://yespress.io/liquid-ai))
- GPU serving via vLLM, SGLang, Baseten, Modal, OpenRouter ([liquid.ai](https://www.liquid.ai/models), [liquid.ai](https://www.liquid.ai/api), [yespress.io](https://yespress.io/liquid-ai))

## Customers and markets

- **Customers**: Mercedes-Benz (multi-year partnership for embedded in-car intelligence / MBUX Virtual Assistant), Shopify (multi-year partnership for sub-20ms foundation models in core commerce; 1B+ LFM requests by Jul 2026), MacPaw (long-term partnership for on-device Mac AI via Eney assistant), Insilico Medicine (strategic partnership for drug-discovery LFMs), Qualcomm (partnership for personal AI context on Snapdragon-powered devices), AMD (strategic partner and lead investor; training/deployment on AMD Instinct GPUs), G42 (commercial partnership), Capgemini (enterprise collaboration), Amazon Web Services (Bedrock/Marketplace distribution) ([ai-market-watch.com](https://www.ai-market-watch.com/company/liquid-ai), [cbinsights.com](https://www.cbinsights.com/company/liquid-ai/customers), [linkedin.com](https://www.linkedin.com/posts/okosovan_today-im-beyond-proud-to-share-that-macpaw-activity-7490747194837467137-8Q33))
- **Markets**: Automotive (in-car intelligence). ([mckinsey.com](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-case-for-liquid-foundation-models), [liquid.ai](https://www.liquid.ai/industrial-robotics), [ai-market-watch.com](https://www.ai-market-watch.com/company/liquid-ai))
- **Markets**: Consumer electronics (on-device personal AI). ([mckinsey.com](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-case-for-liquid-foundation-models), [liquid.ai](https://www.liquid.ai/industrial-robotics), [ai-market-watch.com](https://www.ai-market-watch.com/company/liquid-ai))

## Team

- Ramin Hasani, Co-founder and CEO
- Mathias Lechner, Co-founder & CTO, MIT CSAIL Research Affiliate, PhD from ISTA
- Felix Naser, COO
- Jeffrey Li, COO
- Rom Parnichkun, CTO - Liquid AI Japan
- Rich Counts, VP, Sales
- Leigh Pember, VP of GTM
- Jonathan Mears, VP Finance & Accounting
- Paul Pak, Founding ML Engineer

Backed by: AMD, OSS Capital, PagsGroup / Stephen Pagliuca, Duke Capital Partners, Breyer Capital, Tom Preston Werner / GitHub Co-founder, Naval Ravikant, Safar Partners, ISAI Cap Venture / Capgemini VC Fund, Tobias Lütke / Shopify Co-founder, David Siegel / Two Sigma, Chris Prucha / Notion Co-founder

## Latest news

- 2026-10-27 · [Liquid AI’s new LFM3 models put browser agents, voice commands, and image recognition onto enterprise-controlled devices | VentureBeat](https://venturebeat.com/technology/liquid-ais-new-lfm3-models-put-browser-agents-voice-commands-and-image-recognition-onto-enterprise-controlled-devices) (venturebeat.com)
- 2024-10-23 · [Liquid AI Is Redesigning the Neural Network | WIRED](https://www.wired.com/story/liquid-ai-redesigning-neural-network/) (wired.com)
- 2023-12-06 · [Liquid AI, a new MIT spinoff, wants to build an entirely new type of AI | TechCrunch](https://techcrunch.com/2023/12/06/liquid-ai-a-new-mit-spinoff-wants-to-build-an-entirely-new-type-of-ai/) (techcrunch.com)

## Where to find more

- [This page in Markdown, for agents](https://mad-max.leadbay.ai/company/liquid.ai.md)

## Decision makers

- [Ramin Hasani](https://mad-max.leadbay.ai/person/ramin-hasani--liquid.ai.md), Co-founder and CEO
- [Mathias Lechner](https://mad-max.leadbay.ai/person/mathias-lechner--liquid.ai.md), Co-founder & CTO, MIT CSAIL Research Affiliate, PhD from ISTA
- [Felix Naser](https://mad-max.leadbay.ai/person/felix-naser--liquid.ai.md), COO
- [Jeffrey Li](https://mad-max.leadbay.ai/person/jeffrey-li--liquid.ai.md), COO
- [Rom Parnichkun](https://mad-max.leadbay.ai/person/rom-parnichkun--liquid.ai.md), CTO - Liquid AI Japan
- [Rich Counts](https://mad-max.leadbay.ai/person/rich-counts--liquid.ai.md), VP, Sales
- [Leigh Pember](https://mad-max.leadbay.ai/person/leigh-pember--liquid.ai.md), VP of GTM
- [Jonathan Mears](https://mad-max.leadbay.ai/person/jonathan-mears--liquid.ai.md), VP Finance & Accounting

Work emails and direct phones come with a Leadbay account: https://wow.leadbay.ai/?register=true

Where it comes from: 23 sites read (liquid.ai, exa.ai, beliquid.ai, bloomberg.com, yespress.io, everydev.ai, en.wikipedia.org, venturebeat.com, …).

## Best angle

2026-10-13 · Product: Introducing 𝙻𝙵𝙼𝟹: our new generation of Liquid Foundation Models (LFMs), built for specialized intelligence. (3 sources)

## Best way in

- Message on X: The company's own account

## Signals: what happened

- 2026-11-03 · Headcount: Headcount up 67% in a year (1 source: [A company database](https://www.liquid.ai/))
- 2026-10-27 · News: Liquid AI’s new LFM3 models put browser agents, voice commands, and image recognition onto enterprise-controlled devices (1 source: [venturebeat.com](https://venturebeat.com/technology/liquid-ais-new-lfm3-models-put-browser-agents-voice-commands-and-image-recognition-onto-enterprise-controlled-devices))
- 2026-10-27 · Partnership: Liquid AI introduced LFM3, a family of six compact AI models (language, vision, speech, structured decisions) with 1.7B-parameter foundation, designed for enterprise-controlled devices; Mercedes-Benz and Shopify were pre-release testing partners (4 sources: [venturebeat.com](https://venturebeat.com/technology/liquid-ais-new-lfm3-models-put-browser-agents-voice-commands-and-image-recognition-onto-enterprise-controlled-devices), [liquid.ai](https://www.liquid.ai/news), [liquid.ai](https://www.liquid.ai/blog/d1-decision-model))
- 2026-10-19 · News: Building a model that fits on a phone or laptop makes pre-training harder. With fewer parameters, quality has to come from better architecture and better data, not just more compute. Our pre-training … (1 source: [linkedin.com](https://www.linkedin.com/posts/liquid-ai-inc_building-a-model-that-fits-on-a-phone-or-activity-7517961204712148992-P7Hk))
- 2026-10-15 · News: Today, we are releasing 𝚍𝟷-𝚞𝚕𝚝𝚛𝚊, our most capable decision model yet. Built for hard decision problems, it brings stronger decision quality to complex workflows involving text, images, or both. (1 source: [linkedin.com](https://www.linkedin.com/posts/liquid-ai-inc_today-we-are-releasing-%F0%9D%9A%8D%F0%9D%9F%B7-%F0%9D%9A%9E%F0%9D%9A%95%F0%9D%9A%9D%F0%9D%9A%9B%F0%9D%9A%8A-activity-7516544502317191169-jkBk))
- 2026-10-13 · Product: Introducing 𝙻𝙵𝙼𝟹: our new generation of Liquid Foundation Models (LFMs), built for specialized intelligence. (3 sources: [linkedin.com](https://www.linkedin.com/posts/liquid-ai-inc_today-we-release-%F0%9D%99%BB%F0%9D%99%B5%F0%9D%99%BC%F0%9D%9F%B8%F0%9D%9F%BB-%F0%9D%9F%B8%F0%9D%9F%BA%F0%9D%99%B1-%F0%9D%99%B0%F0%9D%9F%B8%F0%9D%99%B1-activity-7515821552257470464-dIry), [linkedin.com](https://www.linkedin.com/posts/liquid-ai-inc_today-were-announcing-%F0%9D%99%BB%F0%9D%9A%92%F0%9D%9A%9A%F0%9D%9A%9E%F0%9D%9A%92%F0%9D%9A%8D-%F0%9D%9A%81%F0%9D%99%B8-activity-7519080511085031424-Mand), [linkedin.com](https://www.linkedin.com/posts/liquid-ai-inc_introducing-%F0%9D%99%BB%F0%9D%99%B5%F0%9D%99%BC%F0%9D%9F%B9-our-new-generation-activity-7520894654930821121-Vp1P))
- 2026-10-07 · Expansion: Today we release Open d1: d1-3B and d1-omni-600M, two open-weight models in our d1 decision model family. (1 source: [linkedin.com](https://www.linkedin.com/posts/liquid-ai-inc_today-we-release-open-d1-d1-3b-and-d1-omni-activity-7513644633470459906-kyz7))
- 2026-09-26 · News: Liquid AI (2 sources: [linkedin.com](https://www.linkedin.com/posts/liquid-ai-inc_what-makes-an-on-device-agentic-model-actually-activity-7509602663613980674-DH6e), [en.wikipedia.org](https://en.wikipedia.org/wiki/Liquid_AI))
- 2026-09-03 · Interviews: Liquid AI Takes Intelligence Beyond the Cloud (1 source: [open.spotify.com](https://open.spotify.com/episode/1xewXNowQwwK1DBQ5kkwfw))
- 2026-08-05 · News: MacPaw partnered with Liquid AI to bring on-device AI to millions of Mac users via Eney assistant (4 sources: [venturebeat.com](https://venturebeat.com/technology/liquid-ais-new-lfm3-models-put-browser-agents-voice-commands-and-image-recognition-onto-enterprise-controlled-devices), [liquid.ai](https://www.liquid.ai/news), [liquid.ai](https://www.liquid.ai/blog/d1-decision-model))
- 2026-08-03 · Product: Liquid AI on Substack: Model Launches and Research (3 sources: [linkedin.com](https://www.linkedin.com/posts/liquid-ai-inc_liquid-ai-is-now-on-substack-where-well-activity-7483537779902136320-PB01), [linkedin.com](https://www.linkedin.com/posts/liquid-ai-inc_today-we-release-lfm25-26b-an-agentic-activity-7490406467301179393-a0DU), [linkedin.com](https://www.linkedin.com/posts/liquid-ai-inc_today-we-release-updated-4-bit-checkpoints-activity-7495842140468535297-YoRW))
- 2026-05-15 · Interviews: Building a Multimodal Model Without Multimodal Training Data (2 sources: [youtube.com](https://www.youtube.com/watch?v=Gev82mr79-Y), [youtube.com](https://www.youtube.com/watch?v=N0ujymBVxRM))

## Working on now

- Launched the LFM3 family of multimodal foundation models (text, vision, audio, omni, decision) designed for specialized intelligence on Oct 27, 2026.
- Released RIPL (Real-time Inference Pipeline) to enable sub-50ms low-latency inference for enterprise deployments.
- Collaborating with Qualcomm to optimize Liquid Context for Snapdragon processors.
- Developing efficient, hardware-aware agentic models (e.g., LFM2.5, d1 series) for on-device and edge applications.
- Scaling partnerships with major enterprises including Mercedes-Benz, Shopify, and Insilico Medicine.

From: [Introducing LFM3: Every Modality for Specialized Intelligence | Blog | Liquid AI](https://www.liquid.ai/blog/lfm3), [RIPL: Real-time Intelligence When Every Millisecond Counts | Blog | Liquid AI](https://www.liquid.ai/blog/low-latency), [Liquid AI, in Collaboration with Qualcomm Technologies, Brings Personal AI Context to Devices Powered by Snapdragon | Blog | Liquid AI](https://www.liquid.ai/blog/liquid-context-qualcomm), [Liquid AI’s new LFM3 models put browser agents, voice commands, and image recognition onto enterprise-controlled devices | VentureBeat](https://venturebeat.com/technology/liquid-ais-new-lfm3-models-put-browser-agents-voice-commands-and-image-recognition-onto-enterprise-controlled-devices), [We raised $250M to scale capable and efficient general-purpose AI | Blog | Liquid AI](https://www.liquid.ai/blog/we-raised-250m-to-scale-capable-and-efficient-general-purpose-ai), [Liquid AI - AI Startup Profile | AI Market Watch](https://www.ai-market-watch.com/company/liquid-ai)

## Summary

Liquid AI is an MIT spinoff that develops efficient, on-device foundation models designed for local execution without cloud dependency [1][2]. The company builds models that are compact, adaptive, and optimized for low-latency hardware, targeting enterprises that require data sovereignty, privacy, and performance on edge devices [1][3][2]. The company was founded in 2023 by Ramin Hasani, Mathias Lechner, Alexander Amini, and Daniela Rus [4][2]. Key investors include AMD, OSS Capital, PagsGroup, and various industry leaders [4][2]. Major partners and users include Mercedes-Benz and Shopify [3][5][6]. Liquid AI competes with established generative AI firms like OpenAI and Anthropic by offering models that prioritize hardware co-design and efficiency over traditional massive, cloud-bound architectures [7][2][8]. As of October 2026, the company launched its LFM3 model family and the RIPL inference engine, which enables sub-50 ms latency for real-time applications [3][6]. The company offers free access for small-scale use, with paid tiers for larger deployments [2].

[1]: https://www.liquid.ai/company "Company | Liquid AI"
[2]: https://en.wikipedia.org/wiki/Liquid_AI "Liquid AI"
[3]: https://venturebeat.com/technology/liquid-ais-new-lfm3-models-put-browser-agents-voice-commands-and-image-recognition-onto-enterprise-controlled-devices "Liquid AI’s new LFM3 models put browser agents, voice commands, and image recognition onto local devices | VentureBeat"
[4]: https://www.liquid.ai/blog/new-generation-of-ai-models-from-first-principles "Liquid AI: A New Generation of AI Models from First Principles | Blog | Liquid AI"
[5]: https://www.liquid.ai/press/liquid-ai-and-mercedes-benz-partner-to-scale-embedded-in-car-intelligence "Liquid AI and Mercedes-Benz partner to scale embedded in-car intelligence | Press | Liquid AI"
[6]: https://www.marktechpost.com/2026/10/22/liquid-ai-ripl-real-time-inference-engine/ "Liquid AI Launches RIPL, a Real-Time Inference Engine for Sub-50 ms LFMs - MarkTechPost"
[7]: https://www.bizjournals.com/boston/news/2026/08/03/liquidai-how-does-it-work.html "Why this Boston AI unicorn is betting on a different business model than OpenAI's - Boston Business Journal"
[8]: https://techcrunch.com/2023/12/06/liquid-ai-a-new-mit-spinoff-wants-to-build-an-entirely-new-type-of-ai/ "Liquid AI, a new MIT spinoff, wants to build an entirely new type of AI | TechCrunch"

## Sources

- [Liquid AI's website](https://liquid.ai)
- [Company record, via Exa](https://www.liquid.ai/)
- [Introducing LFM3: Every Modality for Specialized Intelligence | Blog | Liquid AI](https://www.liquid.ai/blog/lfm3)
- [RIPL: Real-time Intelligence When Every Millisecond Counts | Blog | Liquid AI](https://www.liquid.ai/blog/low-latency)
- [Liquid AI, in Collaboration with Qualcomm Technologies, Brings Personal AI Context to Devices Powered by Snapdragon | Blog | Liquid AI](https://www.liquid.ai/blog/liquid-context-qualcomm)
- [We raised $250M to scale capable and efficient general-purpose AI | Blog | Liquid AI](https://www.liquid.ai/blog/we-raised-250m-to-scale-capable-and-efficient-general-purpose-ai)
- [Liquid AI - AI Startup Profile | AI Market Watch](https://www.ai-market-watch.com/company/liquid-ai)
- [Liquid AI Raising $250 Million to Build AI Inspired by Worm Brains - Bloomberg](https://www.bloomberg.com/news/articles/2024-12-13/liquid-ai-raising-250-million-to-build-ai-inspired-by-worm-brains)
- [Team | Liquid AI](https://www.liquid.ai/team)
- [Ramin Hasani | MIT CSAIL](https://www.csail.mit.edu/person/ramin-hasani)
- [Daniela Rus receives Bavarian Minister-President's High-Tech Prize | MIT CSAIL](https://www.csail.mit.edu/news/daniela-rus-receives-bavarian-minister-presidents-high-tech-prize)