TL;DR
Anthropic’s Claude Fable 5 is the company’s highest-scoring public model on aggregator benchmarks, but it costs double Opus 4.8. The confirmed price is clear; the public productivity case rests mainly on one unaudited Stripe story cited by Anthropic.
Anthropic’s Claude Fable 5, now listed as generally available after a June 9, 2026 launch, is priced at $10 per million input tokens and $50 per million output tokens, double Claude Opus 4.8, while Artificial Analysis reports a 5.7% Intelligence Index gain over that lower-priced model.
The pricing is confirmed in Anthropic’s launch post, its platform pricing documents and the Fable product page, according to the source material. It places Fable 5 above Opus 4.8 at $5/$25 and well above Sonnet 4.6 at about $3/$15 per million input and output tokens.
The effective cost can fall when buyers use prompt caching, which carries a 90% cache-read discount, or the batch tier at $5/$25. A blended scenario using a 7:2:1 cache-hit-to-input-to-output ratio works out to $7.70 per million tokens, while US-only inference is listed at a 1.1x multiplier.
On performance, Artificial Analysis reports an Intelligence Index score of 64.9 for Fable 5 versus 61.4 for Opus 4.8. In GDPval-AA, a knowledge-work Elo benchmark, the reported scores are 1,932 versus 1,890, a narrower average gap than the price difference suggests.
Budget Pressure for AI Buyers
The confirmed numbers create a direct procurement question: whether twice the list price is justified by a 5.7% aggregate benchmark lead. For teams running large volumes of tokens, even small changes in routing can shift 2026 AI budgets by meaningful amounts.
The case is not one-sided. The source material says Fable 5 set records in 5 of 10 sub-benchmarks, meaning the premium may be easier to defend for specialized work where those gains appear. For broader workloads, the public data points instead to a thin average advantage over a model priced at half the rate.
The Hidden Cost of AI: A Practical Handbook for Token Optimisation: A Practical Handbook for Token and Credit Optimisation in Production LLM Systems (AI Engineering Series)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Fable’s Place in Claude Lineup
Fable 5 is described as Anthropic’s most powerful public model and the most expensive model in its active Claude lineup. The source material says Fable 5 and Mythos 5 use the same underlying weights and carry the same price, with the difference lying in safeguards.
The launch narrative includes one enterprise productivity example. Anthropic’s launch post credits Stripe with using Fable 5 on a 50-million-line Ruby codebase for a migration completed in a day, which the quoted customer account said would otherwise have taken a team more than two months by hand. That statement is a customer claim cited by Anthropic, not an audited productivity study.
“would otherwise have taken a whole team over two months by hand”
— Anthropic launch post, citing Stripe
Stripe Story Leaves Gaps
It is not yet clear how much Fable 5 improves enterprise productivity across ordinary knowledge-work settings. The source material says it found zero controlled human-baseline productivity studies and only one unaudited enterprise proof point tied to Stripe.
Some comparisons also remain less firm. The claim that Fable 5’s input price is roughly 2x GPT-5.5 rests on a single secondary source, according to the source material, so that comparison should be treated as directional. It is also unknown whether Fable 5’s stronger sub-benchmark results map cleanly to specific buyer workloads.
Procurement Tests Come Next
Enterprise buyers are likely to test Fable 5 against Opus 4.8 on their own tasks before committing to broad usage. The next useful evidence would be audited customer studies, workload-level benchmark results and pilot data measuring human review time, token cost and output quality.
Key Questions
What changed with Claude Fable 5?
Claude Fable 5 is now listed as generally available after a June 2026 launch, with pricing of $10 per million input tokens and $50 per million output tokens.
Is Fable 5 twice as intelligent as Opus 4.8?
No public benchmark in the source material supports that. Artificial Analysis reports a 5.7% Intelligence Index gain over Opus 4.8, which is priced at half Fable 5’s list rate.
Why might companies still pay for Fable 5?
The premium may make sense for workloads where Fable 5’s sub-benchmark strengths match the task. Buyers would need their own pilots to show whether those gains offset the higher price.
What evidence is missing?
The missing piece is controlled human-baseline productivity data. The public case now rests mainly on one Stripe customer story cited by Anthropic.
Are Fable 5 and Mythos 5 priced differently?
No. The source material says Fable 5 and Mythos 5 use the same underlying weights and carry identical pricing, with differences in safeguards.
Source: Thorsten Meyer AI
