Liquid AI's banner

Liquid AI

@liquidai • 36,044 subscribers

Build efficient general-purpose AI at every scale.

Shorts

Today we introduce Liquid Labs, our advanced research unit, with the goal of understanding and building efficient and adaptive intelligence systems. Liquid Labs consolidates our existing research efforts at Liquid across architecture of foundation models, multimodality, training, data, and inference. The lab also will be home to new frontier research work across the broad range of foundation model build-up stack. Read the full announcement: We are hiring: Also find us at NeurIPS 2025 exhibition hall! 🚀

Today we introduce Liquid Labs, our advanced research unit, with the goal of understanding and building efficient and adaptive intelligence systems. Liquid Labs consolidates our existing research efforts at Liquid across architecture of foundation models, multimodality, training, data, and inference. The lab also will be home to new frontier research work across the broad range of foundation model build-up stack. Read the full announcement: We are hiring: Also find us at NeurIPS 2025 exhibition hall! 🚀

42,942 просмотров

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

Today we release LFM2.5-Encoder-230M and LFM2.5-Encoder-350M: bidirectional encoders that stay fast at long context, even on CPU. > LFM2.5-Encoder-230M: about 3.7x faster than ModernBERT-base on CPU at 8,192 tokens. Under 30s per forward pass, versus over a minute and a half. > LFM2.5-Encoder-350M: 4th of 14 models on GLUE, SuperGLUE, and multilingual classification, behind only three larger models, one of them nearly 10x its size. 🧵

Today we release LFM2.5-Encoder-230M and LFM2.5-Encoder-350M: bidirectional encoders that stay fast at long context, even on CPU. > LFM2.5-Encoder-230M: about 3.7x faster than ModernBERT-base on CPU at 8,192 tokens. Under 30s per forward pass, versus over a minute and a half. > LFM2.5-Encoder-350M: 4th of 14 models on GLUE, SuperGLUE, and multilingual classification, behind only three larger models, one of them nearly 10x its size. 🧵

198,990 просмотров • 1 месяц назад

Storing too many tools in your context window increases latency and can lead to wrong tool selection. In this demo, we used LFM2.5-ColBERT-350M as a filter to only select the five most relevant tools among 151 options. It's fast and reliable, even without any specific fine-tuning. Try the demo on Hugging Face! And learn more on our blog:

Storing too many tools in your context window increases latency and can lead to wrong tool selection. In this demo, we used LFM2.5-ColBERT-350M as a filter to only select the five most relevant tools among 151 options. It's fast and reliable, even without any specific fine-tuning. Try the demo on Hugging Face! And learn more on our blog:

33,771 просмотров • 3 месяцев назад

a vision language model too fast for human eyes! kudos Xenova 🐐

a vision language model too fast for human eyes! kudos Xenova 🐐

50,495 просмотров • 6 месяцев назад

Check out our demos using LFM2.5-VL-3B, our latest lightweight, vision-language model that reads screens, documents, and the physical world. First up: LFM2.5-VL-3B running fully on-device in the browser with WebGPU to understand a document page. The model parses the entire layout in one pass and returns regions and labels that the interface renders as an overlay. The demo highlights OCR and layout understanding for visually structured content such as forms, reports, receipts, and other documents. 🧵

Check out our demos using LFM2.5-VL-3B, our latest lightweight, vision-language model that reads screens, documents, and the physical world. First up: LFM2.5-VL-3B running fully on-device in the browser with WebGPU to understand a document page. The model parses the entire layout in one pass and returns regions and labels that the interface renders as an overlay. The demo highlights OCR and layout understanding for visually structured content such as forms, reports, receipts, and other documents. 🧵

12,613 просмотров • 1 месяц назад

Building a model is just the start. Post-training makes it useful. Our CTO Mathias Lechner (Mathias Lechner) sits down for a conversation with Maxime Labonne (Maxime Labonne), our head of post-training, on the pipeline that takes a base model from autocomplete to something that can reason and follow instructions.

Building a model is just the start. Post-training makes it useful. Our CTO Mathias Lechner (Mathias Lechner) sits down for a conversation with Maxime Labonne (Maxime Labonne), our head of post-training, on the pipeline that takes a base model from autocomplete to something that can reason and follow instructions.

26,121 просмотров • 4 месяцев назад

Meet Liquid ShieldFlow. An on-device privacy layer powered by a device-native Liquid Foundation Model, Liquid ShieldFlow redacts sensitive data before it ever leaves your machine. No GPU needed and light on memory. It runs on almost any PC, locally, in real time. ShieldFlow was featured yesterday at Microsoft Build for Foundry Local. It also ran live on AMD laptops at 𝐂𝐎𝐌𝐏𝐔𝐓𝐄𝐗 𝐓𝐀𝐈𝐏𝐄𝐈. Request your early access here to ShieldFlow here:

Meet Liquid ShieldFlow. An on-device privacy layer powered by a device-native Liquid Foundation Model, Liquid ShieldFlow redacts sensitive data before it ever leaves your machine. No GPU needed and light on memory. It runs on almost any PC, locally, in real time. ShieldFlow was featured yesterday at Microsoft Build for Foundry Local. It also ran live on AMD laptops at 𝐂𝐎𝐌𝐏𝐔𝐓𝐄𝐗 𝐓𝐀𝐈𝐏𝐄𝐈. Request your early access here to ShieldFlow here:

20,993 просмотров • 3 месяцев назад

As an early look at ongoing work, we deployed LFM2.5-230M on a Unitree G1, running entirely on-device on its onboard NVIDIA Jetson Orin. The model acts as a skill-selection layer, taking in natural-language instructions and decomposing them into sequences of tool calls. After a quick fine-tune, "Hold still for 2s, walk forward at 1 m/s for 3 m, hold a one-leg kneel for 5s, walk back at 0.5 m/s for 3 m" becomes a structured multi-step plan automatically. (3/n)

As an early look at ongoing work, we deployed LFM2.5-230M on a Unitree G1, running entirely on-device on its onboard NVIDIA Jetson Orin. The model acts as a skill-selection layer, taking in natural-language instructions and decomposing them into sequences of tool calls. After a quick fine-tune, "Hold still for 2s, walk forward at 1 m/s for 3 m, hold a one-leg kneel for 5s, walk back at 0.5 m/s for 3 m" becomes a structured multi-step plan automatically. (3/n)

17,313 просмотров • 2 месяцев назад

The bottleneck in LLM inference isn't compute. It's how fast you can move the weights. Our CTO Mathias Lechner, Mathias Lechner, joins Piotr Mazurek, Piotr Mazurek (in SF 🌉), from our inference team, to discuss what actually limits token throughput and how we're optimizing for it.

The bottleneck in LLM inference isn't compute. It's how fast you can move the weights. Our CTO Mathias Lechner, Mathias Lechner, joins Piotr Mazurek, Piotr Mazurek (in SF 🌉), from our inference team, to discuss what actually limits token throughput and how we're optimizing for it.

21,393 просмотров • 3 месяцев назад

Training LFMs at scale means solving parallelism across every layer of the architecture. And not all layers are the same. Our CTO Mathias Lechner (Mathias Lechner) sits down with Liquid's founding engineer Paul Pak (Paul Pak) to talk training infrastructure: Data, tensor, pipeline, expert, and context parallelism, and how they make context parallelism work across hybrid architectures with both attention and convolution operators.

Training LFMs at scale means solving parallelism across every layer of the architecture. And not all layers are the same. Our CTO Mathias Lechner (Mathias Lechner) sits down with Liquid's founding engineer Paul Pak (Paul Pak) to talk training infrastructure: Data, tensor, pipeline, expert, and context parallelism, and how they make context parallelism work across hybrid architectures with both attention and convolution operators.

17,908 просмотров • 3 месяцев назад

Most multimodal systems need data that combines every modality together. Hard to get, expensive to build. Our CTO Mathias Lechner, Mathias Lechner, sits down with Saniya Karwa, Saniya Karwa, from our multimodal research team to talk about building a mode that handles text, audio, and image, and why you might not need as much combined training data as you think.

Most multimodal systems need data that combines every modality together. Hard to get, expensive to build. Our CTO Mathias Lechner, Mathias Lechner, sits down with Saniya Karwa, Saniya Karwa, from our multimodal research team to talk about building a mode that handles text, audio, and image, and why you might not need as much combined training data as you think.

15,396 просмотров • 3 месяцев назад

Well said. AI shouldn't just live in a data center, but it should run on the device you're already using: Whether it's a phone, a car, an appliance, or another everyday device in the real world. That’s what we’re working on at Liquid AI. We appreciate the spotlight on our COO Jeffrey Li (jeffrey li), who joined the AI-Curious Podcast (AI-Curious). Thank you, Jeff Wilser!

Well said. AI shouldn't just live in a data center, but it should run on the device you're already using: Whether it's a phone, a car, an appliance, or another everyday device in the real world. That’s what we’re working on at Liquid AI. We appreciate the spotlight on our COO Jeffrey Li (jeffrey li), who joined the AI-Curious Podcast (AI-Curious). Thank you, Jeff Wilser!

12,995 просмотров • 4 месяцев назад

Tonight at #CES, Liquid’s CEO Ramin Hasani Ramin announced to the world the LFM 2.5 model library. Watch Ramin’s full announcement alongside Lisa Su, CEO of AMD, and learn how we are making intelligence accessible everywhere. Read the full release: LEAP: HF:

Tonight at #CES, Liquid’s CEO Ramin Hasani Ramin announced to the world the LFM 2.5 model library. Watch Ramin’s full announcement alongside Lisa Su, CEO of AMD, and learn how we are making intelligence accessible everywhere. Read the full release: LEAP: HF:

20,422 просмотров • 8 месяцев назад

Больше нет контента для загрузки