Anthropic Releases Two New Claude Models, Making Both Available Across All Major Cloud Platforms

On June 9, 2026, Anthropic released two new versions of its Claude AI model — Claude Fable 5 and Claude Mythos 5 — and made both available the same day across every major cloud platform that enterprises rely on. Anthropic

What Shipped and Where

Claude Fable 5 is available immediately through the Claude API, Anthropic's own Claude Platform, and across Amazon Bedrock (AWS), Google Cloud's Vertex AI, and Microsoft's Foundry. This simultaneous multi-platform release matters because most teams building production systems anchor themselves to one of these big cloud providers. Being available everywhere on day one removes the usual friction: you don't have to wait for your vendor to add support, or rebuild your application to switch providers.

Claude Mythos 5 came out the same day, but with a different path to release. Anthropic had already let a group of developers and enterprise partners try a preview version of Mythos — called Claude Mythos Preview — starting in April 2026 through Project Glasswing. That gave those early partners roughly two months to build with the model and test it on their own work before today's public launch.

The Preview Strategy

The Glasswing preview program is a deliberate choice about how to release a model, and worth understanding. Releasing a "preview" version to a limited group weeks before the public launch isn't new — software companies have done this for decades. But it serves a clear purpose here: Anthropic gets real-world feedback from trusted partners before the model carries millions of requests in production, and those early partners arrive at the public launch with weeks of experience already built in. That head start can matter when you're trying to get something production-ready quickly.

The two-month gap also creates a competitive advantage for early adopters. Developers who used Glasswing have context and optimizations that general users won't have on day one, which tends to let the most sophisticated teams move faster.

A High-Stakes Partnership: Harvey

Among the teams getting early access to Fable 5, the legal AI platform Harvey is notable — and they publicly announced it on launch day. Harvey is now offering Claude Fable 5 to its customers within its own platform.

This partnership signals something important. Legal work — reviewing contracts, analyzing documents, checking whether one clause contradicts another — demands unusual accuracy and trustworthiness. Lawyers stake their reputation on the work they do. The fact that Harvey, a company whose customers hold it accountable for reliability, is an early partner and willing to say so publicly carries weight that no benchmark score can match. It tells you the model is being trusted with real, consequential work.

This also highlights how the competition in enterprise AI is shifting. The underlying model matters, but so does which application — the software you actually use — gets new capabilities first. Harvey's announcement is simultaneously a genuine product milestone and a competitive move against other legal AI platforms.

Two Models for Different Jobs

Releasing Fable 5 and Mythos 5 together invites an obvious question: what are they each for? The naming — Fable and Mythos, both at version 5 — suggests Anthropic is building a family of models, not a single flagship with a cheaper cousin. Think of it the way car manufacturers offer a sedan, an SUV, and a truck: they're built on related platforms but designed for different needs.

Without public benchmarks or pricing details yet, it would be guessing to say exactly what each does better. But structurally, the dual release gives teams a choice within a single vendor's ecosystem. If your application needs one model for high accuracy and another for speed, you can use both Claude versions without juggling separate vendor relationships and integrations.

This idea — offering a range of products tailored to different needs — is not new. Cloud providers like AWS and Google went through the same evolution in the 2010s, eventually learning that customers preferred a coherent family of products they could mix and match across their workloads. The AI model market is following the same path now, with labs offering a spectrum from high-performance to efficient versions released as a family.

The Power of Broad Distribution

The fact that Claude Fable 5 is available on Bedrock, Vertex AI, Microsoft Foundry, and Anthropic's own platform at the same time reflects a business choice: reach matters more than exclusivity. Each platform has strengths — Bedrock fits tightly with AWS security and data tools, Vertex plays well with Google's other ML systems, Foundry aligns with Microsoft's enterprise customers and Azure — and Anthropic is betting that being available everywhere lets more teams use the model.

The practical benefit for you building a production system is straightforward: the model behaves the same regardless of which platform you call it from, and switching costs are low. If you need to move an application from one cloud provider to another, using Claude makes that easier.

What Comes Next

The bigger question is whether Anthropic will use the Glasswing preview approach again for future models. If the two-month preview window helped the company improve Mythos 5 before launch, it could become a standard practice. If it didn't add much value, it was a one-time experiment.

The Harvey partnership also hints at a broader trend. Companies building specialized applications in fields like law, healthcare, and finance — industries where mistakes are expensive and regulated — will likely get early access to new models in exchange for real-world testing and public endorsement. Expect that pattern to expand across other high-stakes verticals.

Both Claude Fable 5 and Claude Mythos 5 are available as of June 9, 2026.