Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on Monday, positioning them for difficult technical work. The company says both names refer to the same underlying model. Fable includes safeguards that block many cybersecurity and biology tasks, while qualified organizations can apply for access to Mythos through Anthropic’s verification programs.
Anthropic says the new versions improve coding, reasoning, and vision, and generate output up to 30 percent faster than Fable 5.1 and Mythos 5.1. The listed price is $8 per million input tokens and $40 per million output tokens. Fable 5.2 is available through Anthropic and on Amazon Web Services, Google Cloud, and Microsoft Azure.
The commercial point is model specialization. Anthropic is separating broad availability from access to higher-risk capabilities, while giving enterprise buyers a new top tier for complex engineering and research. The benchmark results and technical demonstrations come primarily from Anthropic, so customers should test the model on their own workloads before changing production systems.
OpenAI is also making enterprise distribution easier. ChatGPT Enterprise is now available through AWS Marketplace. OpenAI says organizations can buy and manage the product through their existing AWS purchasing and billing relationship, and eligible spend can count toward AWS commitments. Customers receive the same models, features, and release timing available directly from OpenAI.
For technology and finance teams, the appeal is a more familiar procurement route where software purchasing is already standardized around AWS. It also shows that competition among model companies increasingly includes distribution and purchasing channels, not only model performance.
Australia is moving toward tougher rules for frontier AI. The government says it will legislate mandatory safety and security protections for frontier AI companies operating in the country and propose a targeted regulator to enforce them. Proposed obligations include security controls, serious-risk management, independent evaluations, rapid incident reporting, and published safety frameworks.
The direction is clear, but the details are not final. The government says more information will come as it completes consultation and legislative work. Global AI providers should watch the scope of the standards, the penalty structure, and how the regulator defines a frontier system.
In marketing technology, Attentive named co-founder Eric Miao as chief executive and announced general availability of its Model Context Protocol connection. Attentive says marketers can connect supported assistants, including ChatGPT and Claude, to approved Attentive data and workflows. Users can access data, build segments, and create or edit campaigns and journeys through prompts. Marketers control whether changes are enabled and can review customer-facing work before it goes live.
For commerce teams, this is another step toward using general-purpose assistants as an operating layer across customer data and campaign tools. Governance matters here. Access permissions, approval rules, and the quality of customer data will shape whether these workflows save time without creating brand or compliance risk.
And Venmo is rolling out receipt scanning for item-level bill splitting. Users can photograph or upload a receipt, assign items to people, and let Venmo calculate each person’s share of tax and tip before sending payment requests. Venmo says the feature should become widely available in the coming weeks. The user still reviews assignments and can make adjustments before requests are sent.
Watch three things next: independent tests of Anthropic’s new models, the implementation timetable for Australia’s proposed rules, and whether enterprise buyers use existing cloud commitments to accelerate ChatGPT adoption. In commerce, watch how quickly marketers turn assistant connections into governed, repeatable workflows rather than one-off experiments.
That’s your AI Commerce Brief for today. Thanks for listening.
