It was a huge day in AI. Google dropped Gemini 2.5 Pro with insane benchmarks, outperforming models like GPT-4o in reasoning, math, and code. DeepSeek quietly launched a new V3, and it’s MIT licensed with impressive local performance. Meanwhile, OpenAI’s new image model took a massive leap forward—now accurately rendering text, transforming photos, and creating infographics. Here's everything you need to know about the latest in AI. What do you think? Let us know in the comments.

❤️ More about us
Radically better observability stack: betterstack.com/
Written tutorials: betterstack.com/community/
Example projects: github.com/BetterStackHQ

📱 Socials
Twitter: twitter.com/betterstackhq
Instagram: www.instagram.com/betterstackhq/
TikTok: www.tiktok.com/@betterstack
LinkedIn: www.linkedin.com/company/betterstack

📌 Chapters:
0:00 Gemini 2.5 Pro
2:09 DeepSeek V3
2:47 OpenAI Imag