This week in AI: Anthropic's most powerful public model lands, OpenAI deepens its cloud empire, and Washington sets new rules for frontier AI.
1. Anthropic Unleashes Claude Fable 5 — Its Most Capable Public Model Yet
Anthropic released Claude Fable 5 on June 9, 2026, marking the first time the company has made a Mythos-class model available to the general public. Positioned above the Opus tier in capability, Fable 5 achieves approximately 80.3% on the SWE-Bench Pro benchmark and is designed for the kind of long, complex reasoning tasks that previously required restricted enterprise access. A key safety design routes high-risk queries — in areas like cybersecurity, biology, and chemistry — to Claude Opus 4.8 rather than refusing them outright, keeping the model open while managing potential misuse. The model is immediately available on the Claude API, AWS Bedrock, Vertex AI, and Microsoft Foundry, with pricing set at $10 per million input tokens and $50 per million output tokens. Pro and Enterprise users get free access through June 22. A restricted sibling model, Mythos 5, remains available only to a narrow group of cyberdefenders through Project Glasswing.
https://www.anthropic.com/news/claude-fable-5-mythos-5
2. OpenAI and Oracle Forge Enterprise Cloud Alliance
OpenAI and Oracle announced a partnership this week that allows enterprise customers to access OpenAI's frontier AI models — including Codex — directly through their existing Oracle Cloud Infrastructure spending commitments. Businesses can apply eligible Oracle Universal Credits toward OpenAI usage without establishing new contracts or procurement processes, dramatically lowering the barrier to deploying cutting-edge AI in production. The deal is part of a broader $300 billion cloud infrastructure relationship between the two companies that also underpins OpenAI's Stargate data center expansion. The integration is aimed squarely at large enterprises already embedded in Oracle's ecosystem who want to layer AI into data analysis, application development, and workflow automation — without adding yet another vendor relationship.
https://openai.com/index/openai-on-oracle-cloud/
3. Trump Signs AI Executive Order Establishing Voluntary Frontier Model Framework
President Trump signed an executive order on June 2, 2026 titled "Promoting Advanced Artificial Intelligence Innovation and Security," establishing a voluntary framework for developers of the most powerful AI models. The order asks AI companies to give the federal government up to 30 days of early access to "covered frontier models" before broader release, enabling national security benchmarking of advanced cyber capabilities. Crucially, the order avoids any mandatory licensing, preclearance, or permitting requirements, framing itself as pro-innovation while still addressing real national security risks posed by highly capable systems. Agencies are also directed to strengthen federal defenses against AI-enabled cybercrimes, and the administration has set an August 1, 2026 deadline to finalize the voluntary pre-release engagement process.