Anthropic expands Claude access

Anthropic is widening enterprise access to Claude at the same time it is being tested on reliability. The clearest new expansion is around Claude Mythos Preview, Anthropic’s security-focused model for finding software weaknesses. According to reports on June 2 and June 3, Anthropic broadened access to roughly 150 additional organisations across more than 15 countries, including some healthcare users and some organisations in India. That puts Mythos in a more overt enterprise phase, not just a limited technical preview. (letsdatascience.com) A second track is broader workforce deployment. Grant Thornton UK said it plans to roll out Claude to partners and employees across audit, tax, advisory and support functions between June and August as part of a £500 million programme, according to the cited report. That matters because it shows Claude moving from specialist experimentation into day-to-day professional workflows inside a large services firm. (letsdatascience.com) The timing is awkward for Anthropic. The access expansion landed alongside a significant Claude outage reported on the day of Anthropic’s stock market float announcement. The Register said the disruption affected users as the company was marking that market milestone, putting resilience in focus just as Anthropic was trying to show broader enterprise momentum. (theregister.com) For buyers, that combination sharpens a practical question: not whether Claude is available, but whether it is dependable enough for regulated work. Audit, tax, security review and healthcare are the kinds of functions where downtime, latency or inconsistent performance can become an operational issue quickly. The outage does not by itself define Anthropic’s platform, but it gives procurement and risk teams a concrete event to examine. (theregister.com) Anthropic’s own documentation shows why adoption decisions are getting more granular. The company now presents a family of Claude models rather than a single default choice, and its pricing page lays out different price points by model and usage. For enterprises, that means model selection is becoming a budget and governance decision as much as a capability decision. Teams may choose one model for general knowledge work, another for coding or security tasks, and still need fallback plans if service quality slips. (platform.claude.com) That also changes how Claude will be evaluated against rivals. Enterprises are no longer only comparing benchmark performance. They are comparing deployment fit: which model can be approved by security teams, which one has pricing that holds up at scale, and which provider can support production workloads without visible interruptions. Anthropic’s latest moves suggest it is winning more of those evaluations, but the outage ensures reliability will remain part of the same conversation. (theregister.com) The next things to watch are concrete and near term. Grant Thornton UK’s June-to-August rollout will show how Claude performs in a large professional-services deployment, while Anthropic’s model and pricing pages will remain the reference points for how customers map use cases to specific Claude offerings. (letsdatascience.com)