Fable 5 Is Free Until July 7: What to Do With It Before It's Gone

Anthropic just gave every Claude subscriber a free upgrade, and it expires this week.

Fable 5 is the first model in the new Claude 5 family, and it sits in a new tier Anthropic calls Mythos, above Claude Opus in raw capability. It was announced on July 2, 2026, and it is included free with existing Claude subscriptions through July 7. After that, it either becomes a paid tier or disappears from your plan entirely.

Four days is not a lot of time. Here is what is actually worth doing with Fable 5 before the window closes, and why the smartest move is not about the model at all.

What Fable 5 Actually Is

A quick recap of the moving pieces, because two things launched in the same week and it is easy to mix them up.

Claude Sonnet 5 became the default model in Claude Code on June 30, 2026. It ships a native 1M-token context window and introductory pricing through August 31. That one is staying. Fable 5 is different: it is the flagship of the new Mythos class, positioned above Opus, and it is the one with a hard deadline. Anthropic is letting existing subscribers try it free until July 7, per the changelog on yet-another-changelog.ai.

That distinction matters. Sonnet 5 is your daily driver going forward. Fable 5 is a four-day window to borrow the smartest model available and get real, permanent value out of it before it goes away.

GitHub was among the first to put the model through its paces. As Mario Rodriguez, Chief Product Officer at GitHub, put it: "Claude Fable 5 is a real step forward for the developers GitHub serves. In our early testing, it took on complex, long-horizon coding tasks with a level of autonomy and reliability that exceeded previous benchmarks." That is the gap this free window lets you test firsthand, before it closes.

Three Things Worth Doing Before July 7

1. Use the Full Context Window for Planning, Not Just Code

The most clever model I have used to date is genuinely good at leveraging a full context window for prioritization, strategy, and go-to-market thinking, not only code generation. Most people default to using their best model for the hardest coding problem of the week. That is the wrong instinct for a model this good.

Instead, feed it everything: your PRD, your last three retros, your competitor's changelog, your support ticket themes. Then ask it to draft a roadmap prioritization with tradeoffs spelled out, or a first pass at a go-to-market plan. A model that can hold your entire product context at once is rare, and reasoning quality at that scale is exactly what a bigger, smarter model buys you over a faster, cheaper one.

2. Automate Content at Scale, Carefully

Fable 5's reasoning depth makes it tempting to batch-generate content: blog posts, landing page variants, email sequences. Do it, but with a hard rule attached. Automate the volume, never the voice. Obviously AI-written content (generic structure, no specific examples, no opinion) damages a brand faster than it builds one. Use the model to draft and structure at scale, then keep a human pass for judgment, specifics, and the parts only you can say.

3. Let Fable Teach You How to Live Without It

This is the idea worth acting on before anything else. Model access is temporary. What you build with it is not.

Use Fable 5 this week to design the harness you will keep using after it is gone: subagents, custom slash commands, skills, and hooks that encode the judgment you just watched the model apply. Ask it to write the system prompt for a subagent that reviews PRDs the way it just reviewed yours. Ask it to turn its own prioritization framework into a reusable command. Once that harness exists, a cheaper model like Sonnet 5 can run it and get you most of the way to Fable-level output, long after July 7.

The free trial ends. The system you build with it does not.

Fable 5 for Product Managers

Everything above translates directly into PM work, building on what actually changes when a PM tests Fable 5 day to day, and it is exactly where a Mythos-class model earns its keep. Point it at a competitor's product and changelog for a same-day competitive teardown. Feed it a spec and ask it to review it from three angles at once: engineering feasibility, user experience, and business tradeoffs. Hand it a folder of interview notes and get discovery synthesis that would normally take an afternoon.

94% of product professionals now use AI daily or often in their workflow, according to a 2025 Productboard survey of 379 product professionals, with teams reporting roughly four hours saved per task. The gap between PMs who benefit from that shift and PMs who don't is not access to a model. It is knowing what to build with it while you have it.

The Takeaway

Fable 5 is free for four more days. Spend that time on planning and strategy, not just code. Automate content without losing your voice. And above all, use the smartest model you will have access to this year to build the harness that lets a cheaper model do the same job next month. That harness is the actual asset. The free trial was just the occasion to build it.