Fable 5 vs Opus: Claude's Top Two Models Compared

Claude Fable 5 is Anthropic's most capable widely released model at $10 per million input tokens and $50 per million output, while Opus 4.8 sits one tier below at exactly half that price, but as of June 2026 Fable 5 is globally suspended, so Opus 4.8 is the model you can actually run today. In other words, Fable 5 wins on raw ceiling, yet Opus 4.8 wins on availability and cost, which makes Opus 4.8 the practical pick right now.

The quick answer

Opus 4.8 is the model to use right now because Fable 5 is currently suspended and cannot be accessed. Fable 5 launched on June 9, 2026 as Anthropic's most capable widely released model, but within roughly 72 hours it was pulled.

According to reporting, Anthropic globally suspended both Claude Fable 5 and Mythos 5 around June 12, 2026 following a U.S. export-control directive that cited a jailbreak concern. Because the company could not filter access by nationality, it disabled both models worldwide.

That leaves Opus 4.8 as the highest-capability Claude model you can actually deploy today. It is fully available, runs at half the price of Fable 5, and ships as the default high-capability coding model in Claude Code.

Side-by-side comparison

On paper Fable 5 leads on capability tier while matching Opus 4.8 on context and output limits, but the two diverge sharply on price and, critically, on availability. The table below lays out the verified specs side by side.

The headline takeaway: identical context and output ceilings, but Opus 4.8 costs half as much and is the only one of the two you can call right now.

Capability difference

Fable 5 is positioned as Anthropic's most capable widely released model, which is the single reason to prefer it over Opus 4.8 when it is available. Its adaptive thinking is always on, meaning the model reasons through every request rather than treating extended reasoning as an opt-in feature.

Opus 4.8, by contrast, makes thinking optional. That flexibility lets you run fast, cheap responses for routine work and switch on extended reasoning only for the hard problems, which is often the more economical pattern in production.

Fable 5 also has a distinctive safety behavior: it can return a stop_reason of "refusal", declining requests it judges unsafe. For most teams the real-world capability gap is narrower than the tier labels suggest, especially since Opus 4.8 is already strong enough to be the default coding model in Claude Code. I avoid quoting benchmark numbers here because reliable head-to-head scores are not something to invent.

Price gap

Opus 4.8 costs exactly half of Fable 5 on both input and output, which is the clearest financial argument for the lower tier. Fable 5 runs $10 per million input tokens and $50 per million output, while Opus 4.8 runs $5 and $25 respectively.

For a workload that consumes 10 million input and 2 million output tokens in a month, Fable 5 would cost about $200 versus roughly $100 on Opus 4.8. At scale, that 2x difference compounds fast across agents, batch jobs, and long-context pipelines.

There is also a data-handling difference worth pricing in. Opus 4.8 is available under zero data retention, while Fable 5 requires a 30-day retention window, which can matter for regulated or privacy-sensitive deployments.

Which to pick

Right now the choice is simple: use Opus 4.8, because Fable 5 is suspended and cannot be accessed. Even setting the suspension aside, Opus 4.8 is the sensible default for most teams given its half-price economics, optional thinking, and zero-data-retention option.

If and when Fable 5 returns, reserve it for your hardest work, the tasks where you genuinely need the highest capability tier and the always-on adaptive reasoning justifies the doubled token cost. Think deep research, gnarly multi-step reasoning, or your most complex agentic runs.

For everything else, including the bulk of coding, Opus 4.8 is the workhorse. For a deeper comparison of the two tiers, see our Claude Fable 5 vs Opus 4.8 deep dive, and for the full lineup our guide to the best Claude models in 2026.

Limitations and Tradeoffs

The biggest limitation of this comparison is that it is fluid: while Fable 5 is suspended, the practical "winner" is Opus 4.8 by default, not on merit alone. The suspension is reported as temporary rather than a permanent ban, but there is no confirmed return date, so any plan that depends on Fable 5 carries real timing risk.

Opus 4.8 has its own tradeoffs. It sits one tier below Fable 5 on raw capability, and its optional thinking means you have to decide when to enable extended reasoning rather than getting it automatically.

Fable 5's always-on thinking and 30-day data retention requirement are also not free choices. They raise both cost and compliance considerations relative to Opus 4.8's zero-data-retention availability. I have deliberately avoided fabricated benchmark scores; treat tier labels as directional, not precise.

Related Guides

• Claude Fable 5 vs Opus 4.8 (deep dive)
• Why Fable 5 Is Unavailable
• What Is Claude Fable 5?
• Best Claude Models for OpenClaw

FAQ

Is Fable 5 better than Opus 4.8?

Fable 5 is Anthropic's most capable widely released model and sits one tier above Opus 4.8, so on raw capability it leads. But it costs twice as much and is currently suspended, so for any work you need to do today Opus 4.8 is the better practical choice.

Can I use Fable 5 right now?

No. According to reporting, Fable 5 was globally suspended around June 12, 2026 alongside Mythos 5 following a U.S. export-control directive, and it cannot be accessed at the time of writing. The suspension is described as temporary, but no return date has been confirmed.

How much more does Fable 5 cost than Opus?

Exactly double. Fable 5 is $10 per million input tokens and $50 per million output, while Opus 4.8 is $5 and $25 respectively, making Opus 4.8 half the price on both.

Which is better for coding, Fable 5 or Opus 4.8?

Opus 4.8 is the default high-capability coding model in Claude Code and is fully available, which makes it the better coding choice today. Fable 5 may have a higher capability ceiling, but it cannot be used during the suspension, so it is not a practical option for coding right now.

Where can I read the primary sources?

Start with Anthropic's official documentation and independent reporting on the suspension:

Introducing Claude Fable 5 (Anthropic docs)

Claude models overview (Anthropic docs)

Fable and Mythos suspension: security takeaways (Snyk)