Introducing @upstageai's top-performing model, DocParse. It surpasses the performance of recent models such as @MistralAI OCR, @llama_index, and @UnstructuredIO on DP Bench, huggingface.co/datasets/upst…. Our product's users are consistently amazed and continue to use it. We offer our product in various ways, such as API, AWS Marketplace, and On-Prem. Try it out on the Upstage Playground at upstage.ai/products/document….
MULTIMODAL AI
-
Claude 4o and Gemini update long context and image model benchmarks
By
–
we have updated the Long Context and the Image models reading lists in the wake of 4o and gemini's chosen benchmarks today.
-
Sora Update: small artists cooked, end of image absorption
By
–
Honestly, with the Sora update, I understand the artists and AI… Now a large part of the small artists, you are cooked like carrots, finito, bye, hasta luego. Strength to the artists. RIP mass image data absorption….
-
Flux better for photorealism, but impressive for drawings
By
–
No it's DALLE, Flux always works better for photorealism, but here for drawings, especially for writing and details it's insane.
-
AI-Powered Chat Agent for GTC25 Keynote Video Search
By
–
Dive into the #GTC25 keynote with our #AI-powered chat agent, ready to answer your questions about specific moments, what was visually shown or mentioned, and provide summaries with key highlights. Built using Search and Summarization Blueprint (VSS) at
-
Sora better than ChatGPT, I share my tests
By
–
Ok friends, on the Sora website it works much better than on ChatGPT!!!!!! I'll put my experiences under this tweet. Also tell me what you want me to test.
-
Gemini 2.5 Pro will get native images
By
–
Confirmed Gemini 2.5 Pro will get native image output too! No ETA though.
-
ChatGPT 4o integrates new ImageGen for image generation
By
–
If you go to ChatGPT and use the 4o model, you should be able to prompt something like "generate an image of X" and it will generate it with the new ImageGen instead of Dall-e. There was no announcement upon login or anything like that for me.
-
ChatGPT’s Image Generation Enhances Poetry Creation Emotionally
By
–
there’s something really emotionally satisfying about using @ChatGPTapp
’s native image gen on poetry -
OpenAI launches native image output for GPT-4o
By
–
BREAKING 🚨: OpenAI released native image output for GPT-4o! Rolling out to all users at this moment. https://t.co/IVqFhEXIWd pic.twitter.com/ExxbLow3m2
— 🚨 AI News | TestingCatalog (@testingcatalog) 25 mars 2025BREAKING: OpenAI released native image output for GPT-4o! Rolling out to all users at this moment.