Hacker Blogs
Work in progress: 82,651 of 181,233 sites checked so far, more every week · updated 2 Nov 2026Missing a blog? Tell us

Top all-time · #1935

L

Liquid AI

Company blog · liquid.ai/blogai

Liquid AI is the company blog at liquid.ai, writing about ai. Since 2023, 38 different Hacker News readers have linked it 47 times, most often “Designing Loops for Production-Grade Work”. No feed was found, so read it on the site itself.

No feed foundVisit the blog
47mentions?
38HN people?
2023–2026first – last linked?
Links on Hacker News per year · peak 31 in 2026

Liquid AI builds efficient Liquid Foundation Models (LFMs) for on-device, edge, and cloud AI with low latency, privacy, and hardware-aware deployment.

Most-linked posts?

  1. Designing Loops for Production-Grade Work
    3 people · 3 mentions · first linked 2026-08-18
  2. /blog/liquid-neural-networks-research
    3 people · 3 mentions · first linked 2024-09-30
  3. LFM2.5-24B-A2B: Built for fast on-device agents
    3 people · 3 mentions · first linked 2026-10-13
  4. Liquid AI reveals 8B-A1B MoE trained on 38T
    3 people · 3 mentions · first linked 2026-05-28
  5. Liquid AI D1-Ultra
    2 people · 2 mentions · first linked 2026-10-15
  6. Open d1: Edge decision models for text, vision, and audio
    2 people · 2 mentions · first linked 2026-10-08
  7. LFM2-24B-A2B: Scaling Up the LFM2 Architecture
    2 people · 2 mentions · first linked 2026-02-24
  8. Liquid Foundation Models: Our First Series of Generative AI Models
    2 people · 2 mentions · first linked 2024-09-30

What Hacker News says?

  1. tomaskafka · 2026-10-15
    Anyone know how to turn off code mode? I just tried the excellent local https://www.liquid.ai/blog/lfm2-5-24b-a2b model, but the model is trained to talk in bash, while OpenCode stubbornly expects JavaScript, and there seems to be no way to make them work together. The model itself performs excellently in pi.

    Comment linking /blog/lfm2-5-24b-a2b

    on “OpenCode 2.0”
  2. gmays · 2026-10-14

    Submitted LFM2.5-24B-A2B: Built for fast on-device agents as a story · 288 points

    HN discussion
  3. GeorgeTirebiter · 2025-10-17
    I've been impressed with CfCs --- and the stuff Liquid https://www.liquid.ai/research/liquid-neural-networks-resear... is trying to do: model neurons as differential equations, and use CfC methods to alleviate runtime ODE solving, using approximations (or tight integral bounds), so there’s often a tradeoff between approximation error and speed. https://chatgpt.com/share/68f2ada8-06ac-8002-9191-269b0cbba4... The Big…

    Comment linking /research/liquid-neural-networks-research

    on “Intercellular communication in the brain through a dendritic nanotubular network”
  4. jasondavies · 2024-09-30

    Submitted Liquid Foundation Models: Our First Series of Generative AI Models as a story · 183 points

    HN discussion
  5. maxc99 · 2024-09-30
    They explain their architecture in more detail here: https://www.liquid.ai/blog/liquid-neural-networks-research .

    Comment linking /blog/liquid-neural-networks-research

    on “Liquid Foundation Models: Our First Series of Generative AI Models”
  6. Philpax · 2026-10-27

    Submitted Introducing LFM3: Every Modality for Specialized Intelligence as a story · 171 points

    HN discussion
  7. wret-ai-high · 2024-09-30
    no transformers, from their blog post: "Specifically, our analysis informs model building by improving three key aspects: token-mixing structure (how the operator mixes embeddings in the input sequence), channel-mixing structure (how it mixes channel dimensions), and featurization, responsible for modulating computation based on the input context." https://www.liquid.ai/liquid-foundation-models

    Comment linking /liquid-foundation-models

    on “Liquid Foundation Models: Our First Series of Generative AI Models”
  8. acossta · 2026-10-15

    Submitted Liquid AI D1-Ultra as a story · 7 points

    HN discussion
  9. d_burfoot · 2024-09-30
    Well... they at least admit that their model is bad at this [0] > What are Language LFMs not good at today: - Zero-shot code tasks - Precise numerical calculations - Time-sensitive information - Counting r's in the word "Strawberry"! [0] https://www.liquid.ai/liquid-foundation-models#reimagining-m...

    Comment linking /liquid-foundation-models

    on “Liquid Foundation Models: Our First Series of Generative AI Models”
  10. gmays · 2026-10-09

    Submitted Open d1: Edge decision models for text, vision, and audio as a story · 6 points

    HN discussion
  11. a_wild_dandan · 2024-09-30
    More citations from their blog[1] too. [1] https://www.liquid.ai/blog/liquid-neural-networks-research

    Comment linking /blog/liquid-neural-networks-research

    on “Liquid Foundation Models: Our First Series of Generative AI Models”
  12. mseri · 2025-09-23

    Submitted LFM2-2.6B: Redefining Efficiency in Language Models as a story · 5 points

    HN discussion
  13. swyx · 2024-09-30
    this is the "paper": a list of citations https://www.liquid.ai/blog/liquid-neural-networks-research i guess they're not really making an effort to explain how specifically all this comes together to create LFMs. other finds https://x.com/swyx/status/1840794198913794236

    Comment linking /blog/liquid-neural-networks-research

    on “Liquid Foundation Models: Our First Series of Generative AI Models”
  14. Philpax · 2026-10-13

    Submitted LFM2.5-24B-A2B: Built for Fast On-Device Agents as a story · 4 points

    HN discussion
  15. labrador · 2025-10-15
    If the past is any indication, edge models will be the norm. I recently attended a panel at the Computer History Museum where Daniela Rus spoke about her company Liqued AI and it's edge models. Redefining what’s possible on the edge. Liquid’s edge-native stack delivers the fastest AI with zero cloud dependencies — powered by LFM2, built with LEAP, experienced in Apollo. https://www.liquid.ai/

    Comment linking /

    on “Reddit Cofounder Ohanian Says Much of the Internet Is Now Dead”
  16. dataminer · 2026-07-07

    Submitted Reducing Doom Loops with Final Token Preference Optimization as a story · 4 points

    HN discussion
  17. jauntywundrkind · 2026-06-16
    i'd love to get to a point where big models can launch subagents that are fast and local. there's a lot of focus on token rate, but just as much, it feels like there's other latencies that are there with these big batched processing runs, and i think local might have some real wins. Gemma 4 seems already on the right track. lfm2.5-8b-a1b ( https://www.liquid.ai/blog/lfm2-5-8b-a1b ) and DiffusionGemma seem to both be…

    Comment linking /blog/lfm2-5-8b-a1b

    on “Running local models is good now”
  18. nnx · 2026-10-16

    Submitted D1-Ultra: Introducing the Best Decision Model as a story · 3 points

    HN discussion
  19. woadwarrior01 · 2026-03-25
    The only takeaway for me from this is that the Ente is at < $10m/year in revenue. Because the LFM model license has a commercial use threshold of $10m. https://www.liquid.ai/lfm-license

    Comment linking /lfm-license

    on “Local LLM by Ente”
  20. Philpax · 2026-08-24

    Submitted Pipette: A benchmarking suite for on-device intelligence as a story · 3 points

    HN discussion
  21. Alephinitesimal · 2026-08-21

    Submitted LFM2.5-DSpark: Up to 3.2x Faster Inference from H100 to MacB as a story · 3 points

    HN discussion
  22. samstevens · 2026-08-12

    Submitted LFM2.5-VL-3B: A Better and Faster Vision-Language Model for the Edge as a story · 3 points

    HN discussion
  23. BUFU · 2025-08-12

    Submitted LFM2-VL: Efficient Vision-Language Models as a story · 3 points

    HN discussion
  24. ushakov · 2025-01-20

    Submitted LFM-7B: Setting New Standards for Efficient Language Models as a story · 3 points

    HN discussion
  25. pember · 2026-10-22

    Submitted RIPL: Real-time Intelligence When Every Millisecond Counts as a story · 2 points

    HN discussion
  26. pember · 2026-10-08

    Submitted Liquid AI releases Pareto-frontier, multimodal decision model for the edge as a story · 2 points

    HN discussion
  27. pember · 2026-10-05

    Submitted d1: The most capable decision model, now with vision as a story · 2 points

    HN discussion
  28. contextweaver · 2026-08-25

    Submitted Designing Loops for Production-Grade Work as a story · 2 points

    HN discussion
  29. kiyanwang · 2026-08-20

    Submitted Designing Loops for Production-Grade Work as a story · 2 points

    HN discussion
  30. Philpax · 2026-08-04

    Submitted LFM2.5-2.6B: Deploy Agents Everywhere as a story · 2 points

    HN discussion

Search all of HN for liquid.ai

Linked by the same people?