AI News
What's happening in artificial intelligence, explained for everyone
Anthropic launches Claude Fable 5: the public version of its most powerful model with safety guardrails
Anthropic launched Claude Fable 5, the first public version of its Mythos-class model — same power as Mythos 5 but with guardrails blocking responses in cybersecurity, biology and chemistry. Available via API today, subscribers get access progressively through June 22. Price: $10/$50 per million tokens.
Google I/O 2026: Gemini 3.5 and Gemini Omni promise a leap in AI capabilities
At Google I/O 2026, the company announced Gemini 3.5 with improved reasoning and action capabilities, plus Gemini Omni, a completely new family of models. It also introduced better tools to identify AI-generated images, Chrome integration, and visual design changes.
DeepSeek unveils new AI model, breaking reliance on US chips
Chinese startup DeepSeek, which surprised the world last year with an affordable yet powerful AI model, launched its new V4 version. Notably, the model is designed to run on chips manufactured by Huawei, a Chinese company, instead of US Nvidia chips. This represents a major step in China's efforts to develop AI technology without relying on US-made components.
OpenAI launches GPT-5.5: more autonomy, better coding and deeper research
OpenAI announced GPT-5.5, its latest AI model excelling in coding, computer use, and deep research, rolling out to paid subscribers on ChatGPT and Codex.
Google launched Gemma 4: the open source model that beats rivals 20 times its size
Google DeepMind released Gemma 4, its most capable open source AI model family to date. The 31B model ranks #3 globally among open models, outperforming competitors 20 times larger. Best part: it's completely free for commercial use under Apache 2.0 license.
Meta launched Llama 4: what it is and why it matters even if you're not a programmer
Meta released a new version of its free artificial intelligence program. Any company or developer can use it without paying, which changes the game against ChatGPT and Google.
GPT-5.4: the AI that now uses your computer better than you
OpenAI launched GPT-5.4 in March 2026 and achieved something that seemed like science fiction: in standardized tests, the AI operates a computer with more accuracy than the average human. It can browse websites, fill out forms, search for information, and complete tasks on your screen — on its own.
Google launched an AI that can make 8-second videos — here's how it works
By just writing a description, Google's new tool creates a short video. It's not perfect yet, but it already shows where the future of visual content is heading.