AI novinky

Anthropic and OpenAI are pushing the boundaries of AI. We've put their latest models, Claude 4 and GPT-5, head-to-head. Which one understands Czech better, is safer, and offers more value to businesses? We've got the results.

Anthropic just secured its biggest infrastructure deal ever with Google and Broadcom, set to deploy a staggering 3.5 gigawatts of TPU chips starting in 2027. What's more, the company's revenue has exploded from $9 billion to $30 billion in a mere four months!

PrismML has released Bonsai 8B, the first commercially usable 1-bit language model. It's 14x smaller, 8x faster, and 5x more energy-efficient than traditional models — and it runs comfortably on an iPhone or Raspberry Pi without the cloud.

AMD announced Day Zero support for Google Gemma 4 — the new family of open-source AI models (2B to 31B) will run on Instinct accelerators, Radeon GPUs, and Ryzen AI processors from day one, via Ollama, vLLM, or LM Studio.

Researchers at Tufts University showed that robots don't need to consume gigawatts of energy to handle complex tasks. Neuro-symbolic AI trains in 34 minutes instead of 36 hours and achieves 95% accuracy where conventional systems lag at 34%.

German company Agile Robots and Google DeepMind have announced a strategic partnership aimed at deploying Gemini Robotics models into industrial robots. The Agile ONE humanoid enters series production in 2026.

Unemployment in many economies has nearly doubled in just one year — and AI is only getting started. Could we really hit 10% by 2027? And what can you do about it today?

The era of single-purpose AI tools is ending. 'All-in-one' platforms are arriving that handle text, images, video, and audio. Who is setting the tone in 2026 and what does it mean for users worldwide? A comprehensive overview.

The new Claude 4 model generation from Anthropic is rewriting the AI map. It outperforms even the best models from OpenAI. Find out what this means globally, what the pricing looks like, and where you can try it.