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
  • RSS feed
  • Newsletter

© 2026 The Open Weights. An independent publication.

PrivacyTermsSMSAggregated by Claude · curated by humans.
LatestKrea2
KreaText → Image

Krea 2 Arrives as a 12B Open-Weights Image Model

A new text-to-image model ships with a faster Turbo variant and downloadable weights on Hugging Face.

Jun 18, 2026
Major releaseOther
Krea 2

Krea 2 has landed as a 12-billion-parameter text-to-image model, with the weights available to download on Hugging Face. The release pairs the base model with a Turbo variant tuned for faster generation, a now-common split that lets users trade a measure of fidelity for speed when they need quick iterations.

The model is documented in a technical report, and it arrives under a custom license rather than a standard permissive one — a detail worth checking before any commercial use. At 12B parameters, Krea 2 sits in the mid-to-upper range for diffusion-style image generators, large enough to aim for strong prompt adherence and detail without demanding the heaviest hardware.

Why it matters

Open-weights image models remain a competitive and fast-moving space, and a downloadable 12B release gives developers something they can fine-tune, self-host, and build on directly. The two key practical points:

  • A standard model plus a Turbo variant for latency-sensitive workflows.
  • Open weights on Hugging Face, governed by a custom license.

For teams already running open image pipelines, Krea 2 is another option to evaluate against the field. As always with a new release, the real test is how its output quality and speed hold up in hands-on use against established alternatives.

Sources

  • krea/Krea-2-Turbo

    Hugging Face

    Visit

Get the model

Hugging Face

Specs

Parameters12B
ArchitectureKrea2Pipeline
LicenseOTHER
Downloads140.9K
Likes427

Modalities

Text → Image
2 versions — view changelog

0 comments

No comments yet. Be the first to weigh in.

More in Text → Image

Mage-Flow
Microsoft/Text → Image

Microsoft's Mage-Flow packs image editing into 4B

A compact model handles both text-to-image generation and instruction-based edits at native resolution, under a permissive MIT license.

Jul 21, 2026
Unknown/Any-to-Any

Boogu-Image-0.1 Brings Unified Multimodal to Open Source

A new Apache-licensed model family folds bilingual text-to-image generation and instruction editing into one system.

Jul 13, 2026
Qwen-Image-Flash
NVIDIA/Text → Image

NVIDIA distills Qwen-Image for few-step generation

A DMD2-distilled build of Qwen-Image trades sampling steps for speed while keeping the original model's output profile.

Jul 1, 2026