Anthropic AI Updates: April 11, 2026

1. CoreWeave Signs Multi-Year Cloud Deal with Anthropic

Anthropic. CoreWeave announced a multi-year agreement with Anthropic to provide GPU compute capacity for powering Claude AI models at production scale. CoreWeave stock rose 11% on the news, coming one day after the company announced a $21 billion expansion with Meta. With this partnership, nine of the top ten AI model providers now use CoreWeave’s platform. Source

2. Anthropic Revenue Hits $30B ARR, Surpassing OpenAI for the First Time

Anthropic. Multiple outlets reported that Anthropic’s annualized revenue run rate hit $30 billion, surpassing OpenAI’s approximately $25 billion ARR for the first time. The company tripled its revenue in roughly four months, growing from $9 billion ARR at the end of 2025. The milestone reflects explosive enterprise demand for Claude, particularly following the Opus 4.6 and Mythos Preview launches. Source

3. Claude Code v2.1.101 Adds Team Onboarding Command and Security Fixes

Anthropic. Claude Code v2.1.101 shipped with a /team-onboarding command that generates teammate ramp-up guides from local usage patterns, OS CA certificate store trust for enterprise TLS proxies, and auto-creation of default cloud environments for /ultraplan and remote sessions. The release also fixes a command injection vulnerability in POSIX which fallback for LSP binary detection, a memory leak in long sessions, and a hardcoded 5-minute request timeout affecting slow backends. Source

4. Anthropic Temporarily Banned OpenClaw Creator from Claude Access

Anthropic. Anthropic temporarily suspended OpenClaw creator Peter Steinberger’s account for “suspicious activity” following a recent pricing change that ended subscription-based access for third-party harnesses like OpenClaw. The ban was reversed within hours after the post went viral, with an Anthropic engineer stating the company has never intentionally banned anyone for using OpenClaw. Steinberger had previously announced in February he was leaving OpenClaw to join OpenAI. Source

5. Experts Say AI-Driven Hacking Era Already Here Despite Mythos Caution

Anthropic. Fortune reported that while Anthropic restricts Mythos Preview access to give defenders time to harden systems, cybersecurity experts warn that comparable offensive AI capabilities already exist in publicly available models. The article argues that Mythos’s restricted release may be more symbolic than substantively protective, as existing models can already carry out sophisticated cyberattacks in minutes. Source