“Improvements to GPT-4o (April 25, 2025): We’re making additional improvements to GPT-4o, optimizing when it saves memories and enhancing problem-solving capabilities for STEM. We’ve also made subtle changes to the way it responds, making it more proactive and better at guiding
MULTIMODAL AI
-
AvatarFX: Character.ai’s Cutting-Edge Video Generation Technology
By
–
🎞️3/3
— Character.AI (@character_ai) 25 avril 2025
👀Read more about what sets AvatarFX apart from the resthttps://t.co/J3WHiGsYQD pic.twitter.com/8MztsL1CB63/3 Read more about what sets AvatarFX apart from the rest https://
blog.character.ai/avatar-fx-cutt
ing-edge-video-generation-by-character-ai/
… -
AvatarFX Advances Character AI with Fluid Motion Technology
By
–
AvatarFX nails the hard stuff 🎯
— Character.AI (@character_ai) 25 avril 2025
✅Long videos
✅Steady faces
✅Fluid hand + body motion
Your avatar stays in character from start to finish.#AvatarFX #CharacterAI
🎞️1/3 pic.twitter.com/p201Yru9piAvatarFX nails the hard stuff Long videos
Steady faces
Fluid hand + body motion Your avatar stays in character from start to finish. #AvatarFX #CharacterAI
1/3 -
Topaz Labs Image and Video Upscaling Models Now Available on Replicate
By
–
Topaz Labs is an institution in media upscaling.
— Replicate (@replicate) 25 avril 2025
We're introducing their image and video models on Replicate.
Upscale video up to 4k/60fps and image up to 6x resolution.@topazlabshttps://t.co/3kNH7KxCmZ pic.twitter.com/iTel8MhGr0Topaz Labs is an institution in media upscaling. We're introducing their image and video models on Replicate. Upscale video up to 4k/60fps and image up to 6x resolution. @topazlabs http://
replicate.com/topazlabs -
Image Generation Models Lose Detail Through Iterative Processing
By
–
it has a tendency to gradually move towards highly desaturated color palettes. most prompts turn to mush after enough generations
-
Meta’s Token-Shuffle Advances Autoregressive Image Generation
By
–
Even if Yann LeCun isn’t a fan of autoregressive models, Meta’s still pushing out some seriously impressive research using them.
Token-Shuffle: Towards High-Resolution Image Generation with Autoregressive Models
Paper: https://
arxiv.org/pdf/2504.17789 -
OpenAI Launches Advanced Contextual Image Generation Model
By
–
GPT Image Generation
— Perplexity (@perplexity_ai) 25 avril 2025
Try the latest image generation model from OpenAI that supports contextual image generation and editing. pic.twitter.com/mBJ2fY7OsDGPT Image Generation Try the latest image generation model from OpenAI that supports contextual image generation and editing.
-
Perplexity Voice Assistant Launches on iOS Platform
By
–
Perplexity Voice Assistant is now on iOS. https://t.co/P2YsImozQu
— Perplexity (@perplexity_ai) 25 avril 2025Perplexity Voice Assistant is now on iOS.
-
Major AI Product Launches: Voice Assistant, Image Generation, Grok 3
By
–
What We Shipped This Week:
— Perplexity (@perplexity_ai) 25 avril 2025
iOS Voice Assistant
GPT Image Generation
Grok 3 Beta
o4-mini
Here's this week's changelog: pic.twitter.com/ipire8tJp4What We Shipped This Week: iOS Voice Assistant
GPT Image Generation
Grok 3 Beta
o4-mini Here's this week's changelog: -
Generative Art Through Feedback Loops and GPT-Image
By
–
There are many ways of making generative art, but my go-to has always been the feedback loop. Just do one thing over and over on the last output and watch unexpected artifacts emerge. With gpt-image, you can write that whole program in a prompt.
— Gene Kogan (@genekogan) 25 avril 2025
For example: "Make an exact copy… pic.twitter.com/tPCaxEsuASThere are many ways of making generative art, but my go-to has always been the feedback loop. Just do one thing over and over on the last output and watch unexpected artifacts emerge. With gpt-image, you can write that whole program in a prompt. For example: "Make an exact copy