The Open Weights
LatestModelsLeaderboardsCompanies
Subscribe
The Open Weights

The daily record of open-source AI. New model releases, leaderboards, and what's coming next — written for people who ship.

Refreshed every 12 hours

Discover

  • Latest releases
  • New today
  • Trending models

Browse

  • All models
  • Companies
  • Categories
  • Leaderboards

About

  • About
  • Editorial policy
  • Hardware estimates
  • RSS feed
  • llms.txt
  • Newsletter

© 2026 The Open Weights. An independent publication.

PrivacyTermsSMSAggregated by Claude · curated by humans.
LatestLiquidAId1-3B
LiquidAIVision-Language

Liquid AI's d1-3B brings multimodal models to the edge

The new LFM2-based d1-3B is a compact vision-language model aimed at running decisions directly on-device.

Oct 7, 2026
NotableOther
LFM2 d1-3B

Liquid AI has introduced d1-3B, part of a new line of open multimodal "decision" models built on the company's LFM2 architecture and designed to run at the edge. At roughly 3 billion parameters, the model is small enough to deploy on constrained hardware while handling vision-language tasks, according to the company's announcement on Hugging Face.

The pitch here is less about chasing frontier benchmarks and more about placement. By keeping the parameter count modest and targeting on-device inference, d1-3B is positioned for scenarios where sending data to the cloud is impractical — think robotics, cameras, and other latency- or privacy-sensitive applications that need to interpret images and text locally.

Why it matters

The open-weights landscape has been dominated by ever-larger models, but a growing share of real-world demand sits at the opposite end: compact systems that run where the data is generated. A multimodal model in the 1B–7B range that can actually fit on edge hardware fills a practical gap.

  • Multimodal: handles both vision and language inputs
  • Compact: around 3B parameters, sized for edge deployment
  • Open: weights published on Hugging Face under the company's license

Liquid AI frames d1 as a family rather than a one-off, suggesting more variants may follow. For developers building on-device applications, the release offers a lightweight option worth evaluating against existing small multimodal models. Full details are available in Liquid AI's blog post.

Sources

  • Multimodal open d1 decision models for the edge

    Announcement

    Visit
  • LiquidAI/d1-3B

    Hugging Face

    Visit
OlderTII extends the Falcon family into speech recognitionIBM · Speech → Text · 2 months agoNewerNVIDIA's Nemotron-3.5 Bolt targets 24GB GPUsNVIDIA · Vision-Language · 2 months ago

Get the model

Hugging FaceAnnouncement

Specs

Parameters3B
Context window33K tokens
Size6.2 GB
PrecisionBF16
ArchitectureLfm2VlForConditionalGeneration
LicenseOTHER
Downloads18.3K
Likes512

Can you run it?

Runs on just about anything.

  • BF16 (as published)

    8 GB GPU (RTX 4060) · 16 GB laptop or Mac

    7.7 GB
  • 8-bit

    8 GB GPU (RTX 4060) · 16 GB laptop or Mac

    4.8 GB
  • 4-bit

    Any modern laptop · 8 GB RAM

    3.4 GB
Your machine
GGUF builds
  • KV cache sized for 8,192 tokens of context; longer prompts need proportionally more.

Based on Hugging Face weights metadata. Assumes an 8K context and ~1 GB runtime overhead; actual needs vary. How we estimate


Modalities

Vision-Language

The Weekly Weights

Every open release that mattered, one email a week.

0 comments

No comments yet. Be the first to weigh in.

More from LiquidAI

All LiquidAI releases →
LFM3-1.7B
LiquidAI/Text / LLMRuns anywhere

Liquid AI debuts LFM3-1.7B for edge deployment

A compact 1.7B-parameter multilingual model built to run locally and serve as the base for specialized multimodal variants.

Oct 23, 2026
LFM3-Omni-2B
LiquidAI/Any-to-AnyRuns anywhere

Liquid AI's LFM3-Omni-2B brings any-to-any AI to the edge

The 2-billion-parameter model combines vision, audio and speech-to-speech in a package small enough to run on device.

Oct 21, 2026
LFM2.5-24B-A2B
LiquidAI/EmbeddingsConsumer GPU

Liquid AI's LFM2.5 Puts a 24B MoE on Your Device

The new mixture-of-experts model keeps total capacity high while activating only about 2B parameters per token for edge agents.

Oct 12, 2026

More in Vision-Language

All Vision-Language →
SenseNova-Loop-9B
SenseTime/Vision-LanguageRuns on a laptop

SenseTime Releases Looped 9B Multimodal Model

The company's new SenseNova-Loop-9B pairs vision-language capabilities with a recurrent looped-transformer design.

Nov 17, 2026
Hunyuan Hy4-flash
Tencent/Vision-LanguageMulti-GPU server

Tencent releases Hunyuan Hy4-flash multimodal MoE

The lightweight, mixture-of-experts model adds vision and text instruction-following to Tencent's open Hunyuan lineup.

Nov 17, 2026
Inkling 2 Mini
Thinkingmachines/Vision-LanguageConsumer GPU

Thinking Machines releases Inkling 2 Mini, a compact open MoE

The Apache-2.0 multimodal model uses a mixture-of-experts design aimed at running locally rather than in the cloud.

Nov 17, 2026