Image: Pixabay / CC0 (Wikimedia Commons)
In short: In autumn 2026 the major AI providers released new models almost at the same time. OpenAI launched GPT-6 Sol and Luna on 22 September and made GPT-6 available to all ChatGPT users on 7 October. Google followed with Gemini 4 Argon on 13 October. Anthropic released Claude Haiku 5.5 on 7 October and is testing a proactive assistant called Claude Duet. Apple shipped Siri AI with iOS 27, though not in the EU for now. In Europe, the "AI Omnibus" pushed back the deadlines for high-risk AI.
What does AI mean in 2026?
Today "AI" mostly means generative models that create text, images, code or speech: large language models such as GPT-6, Gemini and Claude, image generators and models for robots. The big trend of 2026 is agents, AI systems that carry out several steps on their own, such as using websites, testing code or coordinating appointments.
Key models
| Provider | Model | Date | Focus |
|---|---|---|---|
| OpenAI | GPT-6 Sol / Luna | 22 Sep 2026 | fast and cheap, ChatGPT default |
| OpenAI | GPT-6 Astra | before Sol/Luna | flagship for complex work |
| Anthropic | Claude Haiku 5.5 | 7 Oct 2026 | speed, 1M-token context |
| Gemini 4 Argon | 13 Oct 2026 | coding, knowledge work, cyber defence |
OpenAI: since 7 October, paid ChatGPT users get Sol and Free/Go users get Luna, plus interactive "Intelligent UI" answers. ChatGPT has 1.2 billion weekly users. Details are in GPT-6 explained.
Google: Gemini 4 Argon went first to vetted cyber defenders, then to paid Gemini API users, Google AI Ultra subscribers and enterprises. Introductory pricing is $2 per million input tokens and $10 per million output tokens. A token is a small chunk of text, roughly a short word.
Anthropic: Claude Haiku 5.5 is a fast model with a one-million-token context window. Claude Duet, a proactive agent app, is a research preview for US Pro and Max users.
Apple: iOS 27 and macOS 27 "Golden Gate" arrived on 14 September 2026 with Siri AI, which is not initially available in the EU (see Siri AI in Europe).
Regulation: the AI Act now
Regulation (EU) 2026/1744, the AI Omnibus, entered into force on 27 July 2026 and changed the timeline:
- High-risk AI (Annex III), for example in hiring, credit or education: obligations apply from 2 December 2027.
- High-risk AI in regulated products (Annex I): from 2 August 2028.
- Transparency duties (Article 50): generally in force since 2 August 2026. Users must be able to tell when they are talking to an AI or seeing AI-generated content.
Trends to watch
- agents instead of chatbots
- huge context windows
- on-device AI, for example the A20 Pro in the iPhone 18 Pro
- cybersecurity as a focus
- falling prices
What it means for you
Free AI tools are better than ever. Still, check results, keep sensitive data out and rely on trusted sources for important decisions. See our guide to everyday AI tools and the AI section.
Sources
