We also found that multimodal reasoning capabilities emerge naturally after online RL (even when RL is only done on math & code text data), as long as you start from an initial multimodal model.
MULTIMODAL AI
-
Krea 1 Artist Program Results Showcase
By
–
here are some Krea 1 results from folks on our Artist Program. these ones are from @edbyus
-
Krea 1 Supports 1.5K Native Resolution and 4K Upscaled
By
–
Krea 1 supports 1.5k native resolution and up to 4k upscaled.
-
Krea 1 Introduces Style Reference System for Image Generation
By
–
words can be limiting to describe certain styles.
— KREA AI (@krea_ai) 10 juin 2025
Krea 1 includes a style reference system that will take an image and transfer its style to your prompts. pic.twitter.com/5HDbK1wRZiwords can be limiting to describe certain styles. Krea 1 includes a style reference system that will take an image and transfer its style to your prompts.
-
Krea Introduces First Image Generation Model with Aesthetic Control
By
–
today, we're introducing our first image model: Krea 1.
— KREA AI (@krea_ai) 10 juin 2025
Krea 1 offers superior aesthetic control and image quality. It has a wide range of artistic knowledge and supports style references and custom trainings.
learn how to try our free beta 👇 pic.twitter.com/KdKqaPiCk7today, we're introducing our first image model: Krea 1. Krea 1 offers superior aesthetic control and image quality. It has a wide range of artistic knowledge and supports style references and custom trainings. learn how to try our free beta
-
Visual Prompt Combo Trick Yields Spooky Results
By
–
Tried the visual prompt combo trick (photo + Qwen prompt + Midjourney). Result was spooky good. Like it understood aesthetic intent, not just objects.
-
Image to Kit: AI-Powered 3D Animation Generation Tool Launch
By
–
— Carlos Santana (@DotCSV) 10 juin 2025
We’re thrilled to launch Image to Kit – turn a single image into an animation-ready, modular kit to produce unmatched mesh quality with parts. We leverage inference-time compute to reason, zoom, and predict 3D parts with arbitrary resolution. Available now for premium users
-
AI Advances in Automatic 3D Generation Technology
By
–
No perdamos de vista cómo va mejorando las generación de 3D automática con IA!pic.twitter.com/2g4Rv4vSYG
— Carlos Santana (@DotCSV) 10 juin 2025No perdamos de vista cómo va mejorando las generación de 3D automática con IA!
-
Skywork AI Releases SkyReels-V2 Open-Source Infinite Video Model
By
–
Skywork AI released SkyReels-V2, an open-source infinite-duration AI video model
— Rowan Cheung (@rowancheung) 10 juin 2025
It lets you make cinematic videos of any length for free — straight from your browser
Model weights and inference code now on GitHubpic.twitter.com/cjrTAAjXTvSkywork AI released SkyReels-V2, an open-source infinite-duration AI video model It lets you make cinematic videos of any length for free — straight from your browser Model weights and inference code now on GitHub
-
Manus AI Agent Integrates Google’s Veo 3 for Enhanced Video Generation
By
–
Manus, the AI agent taking on Operator, just integrated Google's Veo 3 model
— Rowan Cheung (@rowancheung) 10 juin 2025
The move enhances Manus' agentic video generation, allowing users to enjoy "extra visual finesse" when creating cinematic content
Available to Basic, Plus, and Pro userspic.twitter.com/oh3vp647uEManus, the AI agent taking on Operator, just integrated Google's Veo 3 model The move enhances Manus' agentic video generation, allowing users to enjoy "extra visual finesse" when creating cinematic content Available to Basic, Plus, and Pro users