Claude Fable 5 Usage Credits: What Changes After July 7, 2026

Quick answer. Through July 7, 2026, Claude Fable 5 is included in Pro, Max, Team, and select Enterprise plans for up to 50% of weekly usage limits. From July 8, Fable 5 no longer draws from subscription limits — it bills through usage credits: pay-as-you-go at standard API rates ($10 / 1M input, $50 / 1M output tokens), enabled and capped from Settings → Usage on claude.ai. Opus 4.8, Sonnet, and Haiku stay inside normal plan limits.

When Anthropic restored Claude Fable 5 on July 1 after the June export-control suspension, it came back with a time limit attached: subscription-included access runs only through July 7. If Fable 5 is wired into your coding agent, your Claude Code sessions, or your daily driver on a Pro or Max plan, this is what actually changes — and what to do before the switch.

What exactly happens on July 8?

Two different billing regimes, one day apart:

  • Through July 7: Fable 5 usage counts against your normal plan limits, capped at 50% of your weekly allowance. No extra cost; hit the cap and Fable 5 pauses until the reset.
  • From July 8: Fable 5 moves entirely to usage credits — consumption-based billing at standard API rates, separate from your subscription. If you haven't enabled credits, Fable 5 requests simply won't run on your plan.

What are usage credits, exactly?

Usage credits are Anthropic's pay-as-you-go layer for paid consumer plans (Pro, Max 5x, Max 20x). Originally built so you could keep working after hitting session limits, they now double as the only route to Fable 5 on those plans. The mechanics, per Anthropic's support docs:

  • Enable them at claude.ai → Settings → Usage. Mobile-app subscribers must enable and purchase on the web app — it can't be done in the iOS/Android apps.
  • Billing is at standard API rates. For Fable 5 that means $10 per million input tokens and $50 per million output tokens — the same as calling it via the API.
  • You can set spend limits, so a runaway agentic session can't silently burn through your balance.
  • They cover Claude.ai and Claude Code alike — combined usage across both surfaces draws from the same pool.
  • Your 5-hour session resets are unaffected, and you can disable credits again at any time.
  • You get a warning first: the UI notifies you when you're approaching limits and confirms before switching to credit billing.

How much will Fable 5 actually cost you?

At $10/$50 per million tokens, Fable 5 is premium-priced — roughly in Opus territory. Ballparks for common workloads:

  • Heavy chat use (say 200K input + 50K output tokens/day): ≈ $4.50/day, ~$135/month on top of your subscription.
  • Agentic coding sessions are the real cost driver — a single long Claude Code session can chew through millions of tokens. An hour of intensive agent work can plausibly cost $5–20 in credits.
  • Light, targeted use (reaching for Fable 5 only on the hardest problems, a few times a day) stays in the low dollars per month.

If you're routinely leaning on Fable 5 for everything, compare that spend against simply parking most work on Opus 4.8 (included in your plan) and escalating selectively — our guide on stretching Claude Code usage limits covers the tactics.

What should you do before July 8?

  1. Decide your default model. If Opus 4.8 handles 90% of your work, make it the default and treat Fable 5 as an escalation path.
  2. Enable usage credits now (web app → Settings → Usage) if you want uninterrupted Fable 5 access — don't discover the cutoff mid-session.
  3. Set a spend limit before pointing any autonomous agent at Fable 5.
  4. Pin a fallback in your tooling. If a pipeline hard-codes Fable 5, make sure it degrades gracefully to Opus 4.8 when credits run out — the same resilience lesson the June suspension taught.

Why is Anthropic doing this?

Anthropic hasn't published a detailed rationale beyond capacity management: Fable 5 returned to a surge of demand after three weeks offline, and it also carries a heavier safety stack — the redeployment added a new classifier that blocks the reported jailbreak technique in over 99% of cases, at the cost of more false positives on routine coding requests. Metering the frontier model through credits while keeping Opus 4.8 unlimited-ish on plans is the pressure valve. Whether Fable 5 returns to subscription limits later remains unannounced.

FAQ

Is Claude Fable 5 free on Pro or Max plans right now?

Through July 7, 2026, yes — it's included for up to 50% of your weekly usage limits. From July 8 it bills through usage credits at API rates instead.

What happens if I don't enable usage credits?

Nothing breaks — you keep your full plan with Opus 4.8, Sonnet, and Haiku. You just won't be able to use Fable 5 on that plan until you enable credits.

Does Fable 5 usage still count against my weekly plan limits after July 7?

No. After the switch, Fable 5 consumption comes out of your credit balance only; your plan limits are reserved for the other models.

Do usage credits work in Claude Code?

Yes — credits apply across Claude.ai conversations and Claude Code terminal usage from the same pool, which is exactly why setting a spend limit before running agents matters.

Is the API affected?

No. Direct API access to Fable 5 was restored July 1 and bills as normal API usage ($10/$50 per 1M tokens). The July 7 change only affects how consumer/team subscription plans access the model.

For the full story of the suspension and restoration, read our Fable 5 restoration explainer and the original Fable 5 launch guide. Choosing between Anthropic's models for coding? See Sonnet 5 vs Opus 4.8.