Anthropic shifts pricing and updates Claude
Anthropic moved Claude Enterprise toward a usage‑based pricing model with a $20-per-user monthly base fee plus additional consumption charges, and the company confirmed a service outage on April 15. The firm also released Claude Opus 4.7, which it says improves coding and vision capabilities—details and pricing are in its docs and recent coverage. ( )
Anthropic has changed Claude Enterprise into a seat-plus-usage product and, a day later, confirmed a service outage across Claude.ai, the application programming interface, and Claude Code. (support.claude.com, anthropic.statuspage.io) On Anthropic’s help page, the company says usage-based Enterprise plans charge a per-user monthly seat fee billed annually, and that the seat fee does not include any model usage. Every token a team uses is billed separately at standard application programming interface rates. (support.claude.com) Anthropic’s public pricing pages say Enterprise is available as a self-serve or sales-assisted plan, while the broader Claude pricing page lists Team and Enterprise alongside consumer tiers such as Pro and Max. The company’s help center says the usage-based Enterprise seat price is the same across self-serve and sales-assisted plans. (claude.com, claude.com, support.claude.com) That structure moves Claude closer to the way cloud software is often sold: a fixed access charge for each employee, then a separate bill for how much computing work they actually run. Anthropic ties that variable bill to token consumption, the unit large language models use to meter text and image processing. (support.claude.com, platform.claude.com) Anthropic’s application programming interface pricing page lists Claude Opus 4.6 at $5 per million input tokens and $25 per million output tokens, with separate lower-priced lines for cache hits and higher-priced lines for cache writes. Anthropic said Claude Opus 4.7 keeps the same pricing as Opus 4.6. (platform.claude.com, anthropic.com) The outage landed on April 15, 2026, as Anthropic’s status page logged “increased errors” beginning at 14:53 Coordinated Universal Time. By 15:40 Coordinated Universal Time, Anthropic said the application programming interface had fully recovered, while Claude.ai and login for Claude Code were still affected. (anthropic.statuspage.io) At 16:29 Coordinated Universal Time, Anthropic posted that “Claude.ai and Platform are down,” and at 17:42 Coordinated Universal Time it said login success rates had stabilized and it was working to fully resolve the issue. Mashable reported that an Anthropic spokesperson said Claude.ai and Cowork were down for most users and some users also had trouble logging into Claude Code. (anthropic.statuspage.io, mashable.com) On April 16, Anthropic released Claude Opus 4.7 and said the new model is generally available across Claude products, its application programming interface, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. The company said Opus 4.7 improves advanced software engineering, follows instructions more precisely, and can process higher-resolution images. (anthropic.com) Anthropic also said Opus 4.7 is “less broadly capable” than Claude Mythos Preview, a model it has kept limited while testing cybersecurity safeguards under Project Glasswing. CNBC reported that Anthropic described Opus 4.7 as its most powerful generally available model even as Mythos remains restricted to a smaller group. (anthropic.com, cnbc.com) The result is a busy 48 hours for Anthropic: a new Enterprise billing model that can rise with customer usage, a visible service disruption on April 15, and a same-week push to sell Opus 4.7 as the version of Claude most businesses can actually buy and deploy. (support.claude.com, anthropic.statuspage.io, anthropic.com)
