We updated our moderation with v7 and it should let 70 percent more things through now because the model itself is much more safe. Still more to come but in the meantime, if you can give us lists of prompts you're having issues with that'll help! Thanks!
GENERATIVE AI
-
V7 Generative AI Platform Releases New Remix and Tile Features
By
–
Midnight release time. V7 now supports remix and tile features. We've upgraded the enhance feature for draft jobs. There's a new prompt bar with dedicated modal toggles. And we've added V7 image feedback buttons to V7 jobs so you can help us improve our image quality. Thanks! <3
-
Alibaba Rumored to Release Qwen 3 Model Soon
By
–
Rumor has it Alibaba might drop Qwen 3 in the next few hours. Thoughts?
-
Llama 4 disappoints on code performance
By
–
Ok les gars, faut que je vous dise un truc.
— VISION IA (@vision_ia) 10 avril 2025
J'avais de grands espoirs pour Llama 4 niveau code… mais après l’avoir comparé à d’autres modèles, y compris l’ancienne version de GPT-4o, la déception est réelle.
C’est loin du niveau annoncé. Rien de bluffant, et certainement pas à… pic.twitter.com/Xy151tFO8DOk guys, I have to tell you something. I had high hopes for Llama 4 in terms of code… but after comparing it to other models, including the older version of GPT-4o, the disappointment is real. It's far from the announced level. Nothing impressive, and certainly not to
-
Gemini Models Demonstrate Powerful 2M Token Long Context Capabilities
By
–
A powerful features of our Gemini models since Gemini 1.5 (including 2.0 and 2.5 models) is their powerful long context capability (2M tokens serving in production). You can process ~1000 pages of text, hrs of video, 10+ hrs ofaudio. Come see @SavinovNikolay discuss this!
-
Universal Visual Grounding for GUI Agents Navigation
By
–
Navigating the Digital World as Humans Do: Universal Visual Grounding for GUI Agents Gou et al.: https://
arxiv.org/abs/2410.05243 #ArtificialIntelligence #DeepLearning #MachineLearning -
Tips for writing effective and collaborative prompts
By
–
—
> Start with “You are an expert…”
> Ask a panel of experts for multiple viewpoints
> Request ratings out of 10 before final answers
> Use "Think step-by-step” for higher quality outputs
> Add "let's solve this together" for collaborative answers Good prompts = creativity
— -
Gemini 2.5 Pro Launches Advanced Deep Research Capability Update
By
–
Big new update to the most powerful Deep Research capability on the market (preferred 2-to-1 over the next best competitor), powered by the most capable AI model: Gemini 2.5 Pro – available now in the @GeminiApp with Gemini Advanced
-
Simple readable code for diffusion model training sampling
By
–
GitHub – yuanchenyang/smalldiffusion: Simple and readable code for training and sampling from diffusion models https://
buff.ly/WDuy3yN
#AI #MachineLearning #DeepLearning #LLMs #DataScience