01What Claude Sonnet 5 Is and How It Works

Claude Sonnet 5 is a rumored new AI model from Anthropic that was recently spotted in test data from a partner provider. It could be a new version of the Sonnet model line, or it might be a different tier like Mythos 5.1 or Mythos 6. The exact name and capabilities are still unconfirmed.

In this episode, Julian Goldie explains how a new slug for Claude Sonnet 5 appeared on a provider's tests. This is similar to what happened in February, which led to the release of Sonnet 4.6. The key takeaway is that model names can be confusing, but the underlying system is what really matters for users.

02Main Rumors and What's New

The biggest news is that Fable 5, Anthropic's most powerful model, was released on June 9 and then taken down on June 12. Shortly after, a new model appeared in test data. Here are the key rumors and confirmed facts:

  • Fable 5 and Mythos 5 both launched on June 9. Mythos 5 was only available for Project Glasswing and Cyber Defenders.
  • Fable 5 was shut down on June 12 for national security reasons. Officials believed someone found a way to bypass it.
  • A new, more capable Mythos version has finished training. It could be called Mythos 5.1, Mythos 6, or something else.
  • Claude Sonnet 5 slug was spotted on a partner provider by AstroPolo and reported by Andrew Curran.
  • Anthropic might keep the new model internal for further development instead of releasing it publicly.

03How to Get Started With Model-Agnostic Systems

To avoid getting stranded when a model is taken down, build on a system instead of a single model. Here's how to set up a flexible setup:

  1. Choose an Agent Operating System (AgentOS) like the one Julian shows in the episode. This system lets you plug in multiple models.
  2. Store your memory, prompts, and agents inside the system. This way they stay the same no matter which model you use.
  3. Connect multiple models such as Claude, Hermes, GLM 5.2, and others. The system should allow easy switching.
  4. Set a default model but keep others as backups. If your default model gets pulled, you can switch in seconds.
  5. Test the system by swapping models. Make sure your workflows still work with a different engine underneath.

04How to Use the System When New Models Drop

When a new model like Claude Sonnet 5 appears, the system approach lets you upgrade instantly. In the episode, Julian shows how users with an AgentOS can change one setting to use the new model. Here's the process:

  • If you own the system, you just point it at the new model. You keep the same memory, prompts, and agents.
  • If you built directly on Fable 5, your tools stop working when the model is removed. You have to rebuild from scratch.
  • Example from the episode: When Fable 5 went dark, system users simply swapped to the next best model and kept shipping.
  • The system always points to the best available model. So every new release actually makes your system better without extra work.

05Results and Comparisons From the Episode

Julian compares two approaches: building on a model versus building on a system. The results are clear:

  • People who built on Fable 5 directly woke up stranded. Their tools just stopped working because the model was gone.
  • People who built on an AgentOS didn't break stride. They changed the engine underneath and kept going.
  • The system gets quietly better every day because it points to the newest best model. It upgrades by changing one setting.
  • In contrast, anyone who rebuilds from scratch with each new model is always behind.

The real lesson: models are rented engines that can be pulled anytime, but the system you own is a durable asset.

06Tips, Limits and Mistakes to Avoid

Based on the episode, here are key tips and common mistakes:

  • Don't hardwire your work to one model. If that model gets taken down, you lose everything. Build on a system instead.
  • Don't chase model names. The name (Sonnet, Mythos, etc.) doesn't matter as much as the system you use. Focus on flexibility.
  • Be ready for changes. Models get pulled, priced up, or switched off by outside forces. A policy change can leave you stranded.
  • Remember that Sonnet sits below Mythos tier. If Claude Sonnet 5 comes out, it won't be as powerful as Fable 5. That's a limit to keep in mind.
  • Check official sources. As of this episode, no Claude Sonnet 5 or new Mythos version is listed on the official Anthropic website.

07Who It Is For and Why It Matters

This episode is for anyone who builds AI applications, uses AI tools, or relies on AI models for their work. It matters because the landscape changes rapidly. Models come and go, but a good system keeps you running.

If you are a developer, business owner, or AI enthusiast, you need to think about durability. The episode shows that the smart move is to own a system that swaps any model in. That way, every new release works in your favor.

You are not a renter of a model. You own the system that controls it.

For those in Julian's AI Profit Boardroom community, the full AgentOS zip file is available with models like Claude, GLM 5.2, and more already wired in. This lets you use new models the moment they are real.