Fable 5 Included Free in Paid Claude Plans Until June 22

claudefable-5.ai News - June 9, 2026

Anthropic is giving paying Claude subscribers a two-week window to use its new flagship model at no additional charge. As part of Tuesday's launch of Claude Fable 5, the company said the model is included in Pro, Max, Team, and seat-based Enterprise plans at no extra cost from June 9 through June 22, 2026.

From June 23, Fable 5 access on subscription plans will require usage credits billed at API rates - $10 per million input tokens and $50 per million output tokens, twice the rate of Claude Opus 4.8. Anthropic frames the change as a capacity decision rather than a permanent pricing tier: the company states an intent to "restore Fable 5 as a standard part of subscription plans" once it has the serving capacity to do so.

How the launch window works

During the free period, Fable 5 usage still counts against each plan's existing limits - and it counts heavily. Anthropic says a Fable 5 conversation consumes roughly twice the usage of the same conversation on Opus 4.8, reflecting the model's higher serving cost. Subscribers on lower-tier plans should expect to hit their caps faster when running Fable 5, particularly on long agentic tasks where the model's always-on adaptive thinking can generate substantial output.

The model is selectable in the model picker across the Claude apps on web, desktop, and mobile. In Claude Code, switching takes a single command: /model fable.

What changes on June 23

After the window closes, subscription plans keep their existing models - Opus 4.8 remains the included flagship - but Fable 5 requests will draw from prepaid usage credits at the same per-token rates as the Claude API. In practice, that turns Fable 5 into a metered add-on for subscribers, similar to how some platforms handle premium model access.

Anthropic's stated position is that this is temporary. The company says it wants Fable 5 to become "a standard part of subscription plans" and describes the credit requirement as a consequence of constrained capacity at launch, not a new pricing philosophy. No date has been given for when included access might return.

The safety fallback toggle

One settings detail worth knowing: Fable 5 ships behind real-time safety classifiers, and Anthropic has acknowledged the system is "stricter than would be ideal... This will be frustrating to some users." A toggle labeled "Switch models when a message is flagged" controls what happens on a false positive. With it enabled, a flagged conversation is handed off to Opus 4.8 rather than refused outright - Anthropic says this affects under 5% of sessions. Users who would rather see an explicit refusal than a silent downgrade can leave it off; critics, including Interconnects' Nathan Lambert, have argued the automatic switch should at minimum come with a notification.

What subscribers should do with the window

For subscribers, the two weeks amount to a free trial of a model that would otherwise cost real money on hard workloads. Some practical guidance, based on early usage reports:

  • Bring your hardest problems now. Fable 5's advantage over Opus 4.8 is largest on the most difficult tasks - large refactors, long-horizon agent runs, deep research. Simple queries will not feel meaningfully different and will burn usage twice as fast.
  • Use it in Claude Code. The most dramatic early results - including Stripe's one-day migration of a 50-million-line codebase and Ethan Mollick's single-prompt video games - came through Claude Code's agentic harness, not chat.
  • Watch your limits. The 2x usage multiplier means Max-plan users have the most headroom for sustained Fable 5 work; Pro users may want to reserve it for tasks where the quality jump matters.
  • Decide whether you want the fallback. If your work touches security, biology, or health data - areas where early users have reported classifier false positives - test how the flagging behaves before relying on the model mid-project.

After June 22, the calculus shifts to cost-per-result. Early evidence cuts both ways: Simon Willison reported spending $110.42 in a single day of API experimentation, while Canva engineers reported Fable 5 producing better results with about half the tokens of Opus 4.8 - which would put its effective cost near parity. For two weeks, at least, subscribers can run that experiment for free.

Related coverage