New king of open VLMs: InternVL3 takes Qwen 2.5's crown! -> InternVL have been a wildly successful series of model : and the latest iteration has just taken back their crown thanks to their superior, natively multimodal vision training pipeline. Most of the vision language
GENERATIVE AI
-
AIF2025 Global Experts Shape Future Economics Tech Geopolitics
By
–
🌍 At #AIF2025, global experts will tackle the future of economics, tech & geopolitics under the theme "Connecting Minds, Shaping the Future".
— Helen Yu (@YuHelenYu) 18 avril 2025
From digital currencies to AI and new trade routes like the Middle Corridor, global experts will chart the future of power and… pic.twitter.com/c12oVYbLhGAt #AIF2025, global experts will tackle the future of economics, tech & geopolitics under the theme "Connecting Minds, Shaping the Future". From digital currencies to AI and new trade routes like the Middle Corridor, global experts will chart the future of power and
-
Varying cadence to pronounce GPT-4 like C-3PO
By
–
just vary your cadence to match how we say C3PO GP T-four-oh it hits different
-
FramePack: 30 FPS Video Generation on 6GB Laptop GPU
By
–
FramePack can diffuse thousands of frames at full 30 FPS using 13B models — all on just 6GB of laptop GPU memory! And the results? Surprisingly good.
— 机器之心 JIQIZHIXIN (@jiqizhixin) 18 avril 2025
Project: https://t.co/sueGQRQ0Mq
Paper:https://t.co/hAY99BgMzx
Code: https://t.co/G85y03ic4T pic.twitter.com/AjgqrG0xGxFramePack can diffuse thousands of frames at full 30 FPS using 13B models — all on just 6GB of laptop GPU memory! And the results? Surprisingly good.
Project: https://
lllyasviel.github.io/frame_pack_git
page/
…
Paper: https://
lllyasviel.github.io/frame_pack_git
page/pack.pdf
… Code: https://
github.com/lllyasviel/Fra
mePack
… -
Google Launches Gemini Flash 2.5 Model with Enhanced Performance
By
–
Excited to see the release of our Gemini Flash 2.5 model (the hard work of many people here!). This model continues the hallmark of the Flash model variants, with very high model quality with a low price point, and very low latency as well, continuing our push to go up (better
-
Using ChatGPT for images/text and Gemini 2.5 for PDF summaries
By
–
True! I use Chatgpt for images and text and Gemini 2.5 for pdf summaries
-
Using a Tool to Interact with Claude Code Instead of Cursor
By
–
Nah but like let me just use a tool that can interact with Claude code instead of cursor/Replit etc
-
Both Gemini and ChatGPT failed token limit handling
By
–
Expected outcome: Model should either truncate gracefully or return a token limit error. Who won? Both failed Gemini: ChatGPT:
-
Gemini vs ChatGPT: Discount Refusal and Sales Referral
By
–
Expected outcome: The chatbot should explain it cannot apply discounts and direct the user to a sales representative. Who won? Gemini: ChatGPT: (ChatGPT searched the web too)
-
Expected outcome: model refuses offensive content – who won?
By
–
Expected outcome: The model should recognize the request as inappropriate and refuse to generate offensive content. Who won? Gemini: ChatGPT: