AI news, 8 October: OpenAI rolls GPT-6 and interactive visuals to every ChatGPT user

OpenAI brings GPT-6 and Intelligent UI to all ChatGPT tiers. OpenAI began a global rollout of GPT-6 in ChatGPT on 7 October, pairing the model with “Intelligent UI,” which embeds interactive charts, tappable buttons, forms, calculators, diagrams and other tools directly into answers. Paid tiers (Plus, Pro, Business, Enterprise) received GPT-6 Sol immediately; free and Go users get GPT-6 Luna starting 8 October. According to OpenAI and reports from The Verge and TechCrunch, the system chooses when visuals help and can generate on-the-spot tools such as bill splitters or adjustable recipes. The company says responses start faster and the model shows stronger resistance to safety bypasses. This marks a major everyday shift for more than a billion weekly users, moving chatbots from text toward mini interactive experiences while distinguishing the announcement from long-term performance claims.

Anthropic launches cheaper, faster Claude Haiku 5.5. Anthropic released Claude Haiku 5.5, its third Claude 5.5-family model in roughly a month, aimed at high-volume, speed-sensitive tasks such as summaries, classification and live support. The company says it costs about 75% less on average than Haiku 4.5 and matches OpenAI’s GPT-6 Luna pricing at $0.10 input / $0.50 output per million tokens for shorter prompts. It adds effort controls and is available across Anthropic’s platforms and major clouds. Reuters and Anthropic note the rapid release cadence reflects intensifying competition among leading labs.

Google signs major nuclear deal to power AI data centers. Google and Constellation Energy announced a 20-year power purchase agreement that will unlock 890 megawatts of additional nuclear capacity through upgrades at 11 existing reactors in Illinois, Pennsylvania and New Jersey, part of a larger roughly 3.6-gigawatt arrangement. Constellation will invest more than $4.3 billion, with first extra power expected in 2028. Reuters and Google’s announcement frame the deal as a response to surging electricity demand from AI infrastructure; Google will also supply AI tools to help optimize plant operations. The move highlights how energy supply has become a central constraint on AI growth.

Mistral previews trillion-parameter European model “Le Chonk”. French lab Mistral launched a public API preview of Mistral Large 4, nicknamed Le Chonk, a roughly 1-trillion-parameter multimodal mixture-of-experts model with about 49 billion active parameters and a long context window. Trained from scratch on European data-center GPUs, it is positioned as a strong open-weight contender, especially in cybersecurity and enterprise tasks, with full weights expected later in October. Mistral and coverage in VentureBeat emphasize European sovereignty and efficiency claims relative to some rivals.

Watchdog rates ChatGPT for Teens an “unacceptable risk”. Common Sense Media’s Youth AI Safety Institute tested OpenAI’s teen experience and gave it the lowest rating after finding parental alerts often failed during extended conversations about self-harm or suicide, and crisis referrals were inconsistent. The group called on OpenAI to restrict the product to adults until fixes are independently verified. OpenAI has disputed that the tests fully reflect real-world safeguards. The report, covered by Axios, The Verge and others, fuels ongoing public debate about AI safety for young users.