AI
OpenAI launches GPT-6 with Intelligent UI as Anthropic slashes Haiku 5.5 prices
OpenAI rolls out GPT-6 with interactive interfaces while Anthropic cuts Haiku 5.5 prices by up to 90 percent, and Google opens its SynthID detector to the public.
This edition was produced with artificial intelligence. Text and voice are generated automatically.
OpenAI rolls out GPT-6 with interactive Intelligent UI for all ChatGPT users
OpenAI is launching GPT-6 with a feature called Intelligent UI, presenting answers as interactive interfaces with graphics, buttons, charts and forms instead of plain text. The format adapts automatically to each question, so comparisons appear side by side and explanations can become interactive charts. Users can have small tools built inside the chat, such as a savings calculator or a game. GPT-6 can respond while still thinking, which OpenAI says cuts wait times by 44 percent and can produce better answers. In internal tests it scored higher than GPT-5.6 on difficult web searches. Paying customers get GPT-6 Sol; free users get GPT-6 Luna. The global rollout starts today for Plus, Pro, Business and Enterprise users, with free and Go users following one day later.
Anthropic releases Claude Haiku 5.5 with steep price cuts across model lines
Anthropic released Claude Haiku 5.5, its fastest and most affordable small model, with benchmark gains over Haiku 4.5 and price cuts across the Haiku and Sonnet lines. Haiku 5.5 targets high-volume, cost-sensitive work such as summarization, database queries, classification and live customer support. It costs about 75 percent less than Haiku 4.5 on average; for prompts up to 100,000 tokens, prices fall by up to 90 percent, while prompts over that threshold cost five times as much. An updated tokenizer consumes slightly more tokens per task, so real-world savings are likely smaller than per-token prices suggest. Haiku 5.5 scores 1,620 on GDPval-AA v2.1 versus 735 for Haiku 4.5, and leads OpenAI’s GPT-6 Luna in every tested category but trails Sonnet 5.5. It is the first Haiku-class model with adjustable reasoning levels and is available on AWS, Google Cloud and Microsoft Azure.
OpenAI releases 722 mathematical papers proving and disproving problems
OpenAI released 722 mathematical papers simultaneously, proving and disproving a range of problems. Academics described the scale as impressive and baffling. AI models’ ability to tackle mathematical problems has increased over months; as recently as 2019, models struggled to achieve a passing grade on GCSE mathematics papers. The previous month, OpenAI solved a problem related to the Navier-Stokes equations of fluid dynamics, described as one of the toughest and most enduring puzzles in mathematics. The mass release marks a significant escalation in AI’s contribution to formal mathematical research and raises questions about how the field will absorb such a volume of machine-generated results.
OpenAI announces 722 mathematical discoveries in one go →
Topics: OpenAI
Google opens SynthID Detector to the public as watermarked media hits 180 billion
Google has opened its SynthID Detector to the public, allowing anyone to check whether an image, video or audio file was created by Google’s AI or its partners. The tool detects invisible watermarks embedded in content from models including Gemini, Veo and Lyria, and supports JPG, PNG, MP4 and MP3 files. SynthID is not a universal AI detector; it only flags content carrying a SynthID watermark, a limitation shared by other detection systems that typically verify only their own provider’s watermarks. OpenAI, Nvidia and Kakao also use SynthID for media, while OpenAI developed its own system for text detection. Apple is expected to join soon, according to Google. SynthID detection is built into Google Search, the Gemini app and Chrome, where Google says one million verification requests arrive daily. Google claims more than 180 billion watermarked images and videos to date.
Google says 180 billion images and videos now carry SynthID watermarks as detector goes public →
Microsoft rebuilds Windows around AI agents with native Copilot and Autopilot
Microsoft unveiled updates to Windows on Wednesday that integrate AI agents and other AI tools directly into the PC operating system. Speaking at San Francisco’s Dogpatch Studios, Chief Executive Satya Nadella said Windows is becoming a hybrid intelligence system, combining cloud-based AI with AI that runs locally through products such as the coding tool GitHub. A central element is integrating a reimagined Copilot AI app into Windows; Copilot will run tasks without accessing the cloud when they do not need heavy compute, according to Jacob Andreou, head of Copilot. The redesigned app becomes a central interface for chatting with AI bots, tapping AI tools, writing code and accessing a new Autopilot feature with agentic capabilities. Autopilot lets users name an agent, assign it a role and goal, and set it on tasks including project updates, tracking deadlines and gathering prep materials. Some features shown are not yet available. Nadella was joined on stage by Nvidia Chief Executive Jensen Huang. The Surface Laptop Ultra uses Nvidia’s RTX Spark superchip and costs $2,599, while the Surface RTX Spark Dev Box costs $5,999. The laptop is available beginning October 16, the box in November. In July, Microsoft said Microsoft 365 Copilot surpassed 30 million paid seats, with net seats added doubling quarter over quarter.
Microsoft shows off new Windows software, revamped for agentic AI →