On the 9th of June 2026, Anthropic released Claude Fable 5, the first model from their most powerful tier to be made available to the general public.
I want to be upfront with you. This is not a tool most businesses will interact with directly, at least not yet.
But what it represents, and what it signals about the direction AI is heading, is worth understanding before it shows up in the tools your teams are already using.
What Is Claude Fable 5?
The Mythos tier, explained simply
The name sounds like it belongs in a sci-fi film, but Mythos is actually the name Anthropic gives to its most powerful class of AI models. Claude Fable 5 sits in that category, and until this week, models at this level had never been made publicly available.
The Mythos tier was first previewed in April 2026 to a small group of vetted security researchers and government partners, and was deliberately held back from general release because of how capable it was, particularly around coding, scientific reasoning and identifying software vulnerabilities.
Fable 5 is the first Mythos-class model Anthropic has made broadly accessible.
It takes that same underlying technology and adds strict safety controls before putting it in front of the public. The most powerful version, Claude Mythos 5, stays restricted to Anthropic’s Glasswing program, made up of vetted partners in security and critical infrastructure.
What it can actually do
Fable 5 supports text, images, documents and tool use, and it works with a context window of up to one million tokens, meaning it can hold an enormous amount of information in a single session.
Early technical reviews put it at the top of current benchmarks for intelligence and coding performance, although some reviewers have also noted frustration with the pricing model and what they see as overly cautious safety filters in certain areas.
For business leaders, that combination of high capability and deliberate constraint is actually the most interesting part of the story.
The Guardrail Design Is the Point
Here is what does not always get said clearly. The reason Anthropic built tiered access and deliberate restrictions into Fable 5 is that the underlying model is genuinely powerful enough to cause harm if it is not handled carefully.
That is not marketing language. It is a considered engineering decision, and it is one worth borrowing from.
If you ask Fable 5 a sensitive question, for example about a cybersecurity exploit or a biological process, it does not answer from its full capability. It routes that question to a less powerful model with tighter controls.
The design principle is simple. Match the level of access to the level of trust and accountability in the person or system asking.
That logic applies at a much smaller scale in your business every day. When your teams use AI tools, whether that is Microsoft 365 Copilot, ChatGPT, Claude or something else, the question is not just “is this useful?” It is “what can this model see, what can it do, and who decided that?”
Most AI risk in business does not come from bad actors. It comes from well meaning people working faster than the guardrails around them. A team member who lets a document AI read a confidential contract is not trying to create a problem, but without the right controls in place, they might create one anyway.
What Changes If You Are Already Using Claude or Copilot?
Claude inside Microsoft 365 is already real
If you are primarily a Microsoft 365 business, Claude is not a separate world. Anthropic’s Claude models are already available inside Microsoft 365 Copilot Chat, which means the capability improvements that flow from Fable 5 will eventually show up in the tools your teams are already using.
The platforms are converging, not competing.
The detail of that integration is laid out in this overview of Claude AI in Microsoft 365 Copilot, including where Claude helps most with research, drafting and analysis, and what that means for governance in a Microsoft centric environment.
Pricing is about to shift
Fable 5 is available at no extra charge to Anthropic’s Pro, Max and enterprise subscribers until around 22 June, after which Anthropic will move to a credit based pricing model.
If your business or any of your teams are paying directly for Claude access, it is worth checking what plan you are on and whether your usage patterns are likely to attract higher costs after that date.
This is also a good time to check whether your AI tool access is being managed centrally or whether individual teams are signing up for their own subscriptions. Shadow AI, where people use tools that IT and leadership are not across, is still one of the most common governance gaps we see.
The patterns are familiar, as described in Managing Copilot agent sprawl, and the fix is usually a mix of better visibility, clearer rules and a small number of well supported tools.
Three Questions Worth Asking Your Team This Week
You do not need to become an AI researcher to respond sensibly to news like this. You just need a short list of questions that help you understand where your business actually sits.
1. Which AI models are your teams currently using, and who approved them? If you cannot answer this confidently, you have a visibility gap. A simple AI inventory is a better starting point than another pilot.
2. What data can those models see? This is the question most businesses skip. It is not just about whether the tool is turned on. It is about whether it can access email, SharePoint, confidential documents or client data without anyone having explicitly thought that through.
3. What would happen if a model gave a wrong or harmful answer? Who would catch it? Who is accountable? In a world where AI is doing more of the drafting, summarising and analysis, accountability for AI outputs needs to sit somewhere clear.
If those three questions feel hard to answer right now, that is genuinely useful information. It means your governance foundations need attention before capability becomes a liability.
Where to Go From Here
The Fable 5 release is a reminder that the AI market is not slowing down. Within the next 12 months, it is reasonable to expect that models at Fable 5’s current capability level will be widely accessible and embedded into the tools your business already relies on. The time to build governance around that is not after it happens.
If you are working out how to think about a multi model AI environment, where Microsoft 365 Copilot sits alongside Claude, ChatGPT and potentially open source tools, this guide on designing an AI operating model for your business is a practical place to start. It walks through how to decide which work belongs on which platform and how to put governance in place across all of them without turning it into a bureaucratic exercise.
If you are at the earlier stage, and you are still asking whether your Microsoft 365 data and security foundations are actually ready for AI, the Microsoft 365 Copilot Readiness Assessment is designed to give you a clear picture of where you stand and what needs to be addressed first.
The businesses getting ahead of this are not waiting for the technology to mature. They are building the conditions to use it well.
About the Author
Carlos Garcia is the Founder and Managing Director of CG TECH, where he leads enterprise digital transformation projects across Australia.
With deep experience in business process automation, Microsoft 365, and AI-powered workplace solutions, Carlos has helped businesses in government, healthcare, and enterprise sectors streamline workflows and improve efficiency.
He holds Microsoft certifications in Power Platform and Azure and regularly shares practical guidance on Copilot readiness, data strategy, and AI adoption.
On the 9th of June 2026, Anthropic released Claude Fable 5, the first model from their most powerful tier to be made available to the general public.
I want to be upfront with you. This is not a tool most businesses will interact with directly, at least not yet.
But what it represents, and what it signals about the direction AI is heading, is worth understanding before it shows up in the tools your teams are already using.
What Is Claude Fable 5?
The Mythos tier, explained simply
The name sounds like it belongs in a sci-fi film, but Mythos is actually the name Anthropic gives to its most powerful class of AI models. Claude Fable 5 sits in that category, and until this week, models at this level had never been made publicly available.
The Mythos tier was first previewed in April 2026 to a small group of vetted security researchers and government partners, and was deliberately held back from general release because of how capable it was, particularly around coding, scientific reasoning and identifying software vulnerabilities.
Fable 5 is the first Mythos-class model Anthropic has made broadly accessible.
It takes that same underlying technology and adds strict safety controls before putting it in front of the public. The most powerful version, Claude Mythos 5, stays restricted to Anthropic’s Glasswing program, made up of vetted partners in security and critical infrastructure.
What it can actually do
Fable 5 supports text, images, documents and tool use, and it works with a context window of up to one million tokens, meaning it can hold an enormous amount of information in a single session.
Early technical reviews put it at the top of current benchmarks for intelligence and coding performance, although some reviewers have also noted frustration with the pricing model and what they see as overly cautious safety filters in certain areas.
For business leaders, that combination of high capability and deliberate constraint is actually the most interesting part of the story.
The Guardrail Design Is the Point
Here is what does not always get said clearly. The reason Anthropic built tiered access and deliberate restrictions into Fable 5 is that the underlying model is genuinely powerful enough to cause harm if it is not handled carefully.
That is not marketing language. It is a considered engineering decision, and it is one worth borrowing from.
If you ask Fable 5 a sensitive question, for example about a cybersecurity exploit or a biological process, it does not answer from its full capability. It routes that question to a less powerful model with tighter controls.
The design principle is simple. Match the level of access to the level of trust and accountability in the person or system asking.
That logic applies at a much smaller scale in your business every day. When your teams use AI tools, whether that is Microsoft 365 Copilot, ChatGPT, Claude or something else, the question is not just “is this useful?” It is “what can this model see, what can it do, and who decided that?”
Most AI risk in business does not come from bad actors. It comes from well meaning people working faster than the guardrails around them. A team member who lets a document AI read a confidential contract is not trying to create a problem, but without the right controls in place, they might create one anyway.
If you want a practical view of what connected controls look like across Copilot, data, identity and security, this guide to unified AI governance with clear guardrails for Copilot and beyond shows how to keep AI use visible and aligned with your risk settings.
What Changes If You Are Already Using Claude or Copilot?
Claude inside Microsoft 365 is already real
If you are primarily a Microsoft 365 business, Claude is not a separate world. Anthropic’s Claude models are already available inside Microsoft 365 Copilot Chat, which means the capability improvements that flow from Fable 5 will eventually show up in the tools your teams are already using.
The platforms are converging, not competing.
The detail of that integration is laid out in this overview of Claude AI in Microsoft 365 Copilot, including where Claude helps most with research, drafting and analysis, and what that means for governance in a Microsoft centric environment.
Pricing is about to shift
Fable 5 is available at no extra charge to Anthropic’s Pro, Max and enterprise subscribers until around 22 June, after which Anthropic will move to a credit based pricing model.
If your business or any of your teams are paying directly for Claude access, it is worth checking what plan you are on and whether your usage patterns are likely to attract higher costs after that date.
This is also a good time to check whether your AI tool access is being managed centrally or whether individual teams are signing up for their own subscriptions. Shadow AI, where people use tools that IT and leadership are not across, is still one of the most common governance gaps we see.
The patterns are familiar, as described in Managing Copilot agent sprawl, and the fix is usually a mix of better visibility, clearer rules and a small number of well supported tools.
Three Questions Worth Asking Your Team This Week
You do not need to become an AI researcher to respond sensibly to news like this. You just need a short list of questions that help you understand where your business actually sits.
1. Which AI models are your teams currently using, and who approved them? If you cannot answer this confidently, you have a visibility gap. A simple AI inventory is a better starting point than another pilot.
2. What data can those models see? This is the question most businesses skip. It is not just about whether the tool is turned on. It is about whether it can access email, SharePoint, confidential documents or client data without anyone having explicitly thought that through.
3. What would happen if a model gave a wrong or harmful answer? Who would catch it? Who is accountable? In a world where AI is doing more of the drafting, summarising and analysis, accountability for AI outputs needs to sit somewhere clear.
If those three questions feel hard to answer right now, that is genuinely useful information. It means your governance foundations need attention before capability becomes a liability.
Where to Go From Here
The Fable 5 release is a reminder that the AI market is not slowing down. Within the next 12 months, it is reasonable to expect that models at Fable 5’s current capability level will be widely accessible and embedded into the tools your business already relies on. The time to build governance around that is not after it happens.
If you are working out how to think about a multi model AI environment, where Microsoft 365 Copilot sits alongside Claude, ChatGPT and potentially open source tools, this guide on designing an AI operating model for your business is a practical place to start. It walks through how to decide which work belongs on which platform and how to put governance in place across all of them without turning it into a bureaucratic exercise.
If you are at the earlier stage, and you are still asking whether your Microsoft 365 data and security foundations are actually ready for AI, the Microsoft 365 Copilot Readiness Assessment is designed to give you a clear picture of where you stand and what needs to be addressed first.
The businesses getting ahead of this are not waiting for the technology to mature. They are building the conditions to use it well.
About the Author
Carlos Garcia is the Founder and Managing Director of CG TECH, where he leads enterprise digital transformation projects across Australia.
With deep experience in business process automation, Microsoft 365, and AI-powered workplace solutions, Carlos has helped businesses in government, healthcare, and enterprise sectors streamline workflows and improve efficiency.
He holds Microsoft certifications in Power Platform and Azure and regularly shares practical guidance on Copilot readiness, data strategy, and AI adoption.
Sources
Recent Posts
Popular Categories
Archives