Open Source
MOSS Image: Open Diffusion Model with JSON Layout Control
Sam Witteveenhigh confidence
Why it matters
- →JSON layout control enables deterministic placement of text and objects, solving the 'randomness' problem in generation.
- →Apache 2.0 licensing allows unrestricted commercial use, lowering barriers for enterprise integration.
- →Compatibility with ComfyUI and SGLang ensures immediate adoption by existing developer workflows.
The OpenMOSS-Team has released MOSS Image, a 5B parameter diffusion transformer licensed under Apache 2.0. The model distinguishes itself by supporting JSON-based layout control, allowing users to specify the position of text and visual elements with precision. It is compatible with major inference engines including ComfyUI, SGLang, and vLLM-Omni. This release provides an open-source alternative for deterministic image composition, addressing a common limitation in generative AI.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1FLUX 3: A New Multimodal Model by Black Forest Labs
Replicate Blog · August 4, 2026 · Background
- 2AI Model Trained to Paint Watercolours
Hugging Face Blog · September 3, 2026 · Background
- 3OpenAI Launches ChatGPT Images 2.5
OpenAI · September 8, 2026 · Background
- 4OpenAI Launches ChatGPT Sketch for Image Creation
The Verge AI · September 8, 2026 · Background
- 5ChatGPT 6 Astra: Hands-on with Multimodal Capabilities
Skill Leap AI · September 9, 2026 · Background
- 6OpenAI Releases ChatGPT Images 2.5 and Navier-Stokes Solution
Matt Wolfe · September 11, 2026 · Background
- 7OpenAI tests visual product ads in ChatGPT image generation
The Verge AI · October 5, 2026 · Background
- 8FLUX.3 Image Klein arrives in Diffusers
Hugging Face Blog · October 15, 2026 · Related
More from Sam Witteveen
More in Open Source
© Lev SelectorOpen Sourcemodels
Mistral Large 4 Released as Open Weights Model
French AI startup Mistral has released its latest large language model, Large 4, with open weights available for download.
Lev Selector
© The Verge AIOpen Sourceother
Akai opens MPC Sample Key firmware to modders
Akai is betting that the community will build better instruments than its engineers can. By open-sourcing the firmware for the budget MPC Sample Key and providing an SDK, they are handing the keys to hackers who have already turned the device into a synth powerhouse using AI-assisted coding. This isn't just about patches; it's a strategic pivot where Akai provides the hardware canvas while the community paints the OS. It signals a shift from closed ecosystems to developer-led innovation in music tech.
The Verge AI
© Hugging Face BlogOpen Sourceother
NVIDIA and AI4Bharat release 2.8T Indic tokens
This is the first major open synthetic dataset for India's 22 constitutional languages, solving the data scarcity that has kept Indic models behind English. By releasing both the corpus and the NeMo Data Designer pipeline, NVIDIA gives builders a reproducible recipe to generate high-quality technical content in native scripts and romanized text. The inclusion of math, code, and QA pairs means models can finally handle complex reasoning in these languages rather than just basic translation. This shifts the bottleneck from data collection to model architecture for the next wave of multilingual AI.
Hugging Face Blog