OpenAI plans to launch GPT-5.6 models next week, including Mini and Pro variants that promise significant performance improvements over current offerings.

The new models will feature a 1.5 million token context window, substantially larger than previous versions. Key enhancements include improved long-horizon coding capabilities and faster Codex response times for developers.

OpenAI is positioning the release with competitive pricing designed to undercut Anthropic, particularly as US regulatory issues impact the availability of Anthropic's Claude Fable 5 model.

Memory and artifacts developments

Perplexity unveiled Brain, a memory system that builds persistent context graphs across tasks, projects, and sources. The system allows AI agents to start with relevant context rather than beginning from scratch, linking every memory to its original source while continuously updating knowledge over time.

Meanwhile, Anthropic launched artifacts support in Claude Code, turning work sessions into live, shareable visual pages. The feature offers version history and privacy controls, currently available in beta for Claude Team and Enterprise customers.

Industry tensions and chip competition

Yann LeCun, Meta's chief AI scientist, criticized Elon Musk's xAI as a "failure" unable to compete with leading AI labs. LeCun highlighted Musk's difficulty hiring top talent due to past workplace issues.

Google is expanding its AI chip business by renting TPU computing power to companies including Anthropic at a data center in Western New York. The move represents Google's push to commercialize its tensor processing units as a rival to Nvidia's dominance in AI hardware.

Midjourney announced an unexpected pivot into medical technology with a full-body ultrasonic scanner that can create 3D body maps in under 60 seconds, compared to 60-90 minutes for traditional MRI scans. The company plans to house the machines in dedicated spas for customer use.