BREAKING : Grok 4 will be SOTA – 35% on HLE, 45% with reasoning
– 87-88% on GPQA – 72-75% on SWE Bench (for Grok 4 Code) * Not official benchmarks, I plotted these based on the leaked scores and results from the web for other models
MULTIMODAL AI
-
Grok 4 SOTA avec des scores élevés
By
–
-
GLM-4.1V-Thinking: Powerful Vision-Language Model for Multimodal Reasoning
By
–
GLM-4.1V-Thinking – a powerful new vision-language model for multimodal reasoning!
From STEM to GUI agents, it outperforms models 8x its size.
Open-source, scalable, and state-of-the-art.
Paper Link: https://
arxiv.org/abs/2507.01006 #AI #VLM #Multimodal #GLM4 #OpenSourceAI -
AI voice capabilities raise questions about deployment
By
–
Je ne me l'explique pas. Elle n'a pourtant pas une bouche de l'emploi…
-
Krea Launches Gen-4 Video Model for Cinematic Scene Generation
By
–
Gen-4 is now available in Krea.
— KREA AI (@krea_ai) 4 juillet 2025
this video model can generate cinematic scenes in less than a minute – works great with Krea 1 images!
try it now. pic.twitter.com/eQ1Tb3MSIQGen-4 is now available in Krea. this video model can generate cinematic scenes in less than a minute – works great with Krea 1 images! try it now.
-
Veo 3 Separate Video Audio Prompting Capabilities
By
–
Good morning. It's interesting you can prompt the video and audio parts of Veo 3 completely differently and it works. pic.twitter.com/TBrikvjWhE
— fofr (@fofrAI) 4 juillet 2025Good morning. It's interesting you can prompt the video and audio parts of Veo 3 completely differently and it works.
-
Freepik CEO explains future of AI-generated content
By
–
Here's a deeper dive of it through Freepik's CEO break down on exactly what this means for you and the future of AI-generated content: https://
x.com/cuenca/status/
1940425849486135339
… -
Unlimited AI Image Generation Across Every Model
By
–
Unlimited images for EVERY model
Yes, EVERY AI model offered: Google Imagen 4
Seedream
ChatGPT-4
Kontext Create freely, experiment endlessly—no restrictions. -
Google Engineer Uses Gemini 2.5 for Creative Video Prompts
By
–
Anna Bortsova, a Google UX Engineer, used Gemini 2.5 Pro to generate text-to-video prompts for Veo 3 and created a series of charming paper engineering stop-motion animations with paper-rustling sounds.
— Google AI (@GoogleAI) 3 juillet 2025
After experimenting in the @GeminiApp to help with prompt writing, she… pic.twitter.com/zZLLo4XYqEAnna Bortsova, a Google UX Engineer, used Gemini 2.5 Pro to generate text-to-video prompts for Veo 3 and created a series of charming paper engineering stop-motion animations with paper-rustling sounds. After experimenting in the @GeminiApp to help with prompt writing, she
-
Voyager SDK 1.3.1 Adds LLM Support and YOLOv11
By
–
Voyager SDK 1.3.1 is out! Adds precompiled LLM support (Llama3, Phi3-mini) YOLOv11 + Attention layer support Native Windows runtime (10/11 + Server 2025) Dynamic fan control + broader temp range New use cases: Super Res, Depth Est, LPR Check out the full
-
AI Generated Image Quality: Visible Flaws in Human Appearance
By
–
Kind looks AI generated – the guy has no jacket whatsoever