TL;DR
Anthropic’s Claude Fable 5 is priced at $10 per million input tokens and $50 per million output tokens, double Claude Opus 4.8, according to the provided pricing record. The central issue is that third-party benchmark gains appear narrower than the price increase, while enterprise productivity evidence remains limited to a single unaudited customer account.
Anthropic’s Claude Fable 5 is now positioned as the company’s most expensive active public model, priced at $10 per million input tokens and $50 per million output tokens, while independent benchmark data cited in the source material shows only a 5.7% Intelligence Index gain over Claude Opus 4.8, a model listed at half the price.
The confirmed pricing cited in the source material places Fable 5 at $10/$50 for input and output tokens, compared with Opus 4.8 at $5/$25 and Sonnet 4.6 at about $3/$15. The same source states that Fable 5 is the most expensive model in Anthropic’s active Claude lineup.
Artificial Analysis is cited as showing Fable 5 at 64.9 on its Intelligence Index, compared with 61.4 for Opus 4.8. On the GDPval-AA knowledge-work benchmark, the source lists 1,932 for Fable 5 and 1,890 for Opus 4.8. Those figures point to a narrower measured performance gap than the model’s list-price gap.
The source material also says Fable 5 supports 90% prompt-caching discounts, batch pricing of $5/$25, and a blended rate estimate of $7.70 per million tokens under a 7:2:1 cache-hit, input and output mix. Those discounts may lower real usage costs, but the reported list price remains the clearest confirmed comparison.
Budget Pressure for AI Buyers
The development matters because enterprise teams are being asked to justify higher AI spending with evidence of better output, faster work, or lower total operating cost. Based on the figures in the source material, Fable 5 costs twice as much as Opus 4.8 while showing a smaller average benchmark gain.
That does not mean Fable 5 lacks value. The same benchmark record cited in the source says Fable 5 was a number-one launch on some aggregate measures and set records in five of 10 sub-benchmarks. For buyers, the practical question is whether those strengths match their highest-value workloads.
The more sensitive issue is productivity proof. The source says the public human-baseline evidence for enterprise gains consists of one unaudited customer story, attributed to Stripe. That leaves finance, procurement and engineering leaders with a gap between documented pricing and verified workplace impact.
AI language model pricing comparison
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
According to the supplied source material, Anthropic’s own launch and pricing materials confirm the Fable 5 list price, while outlets and firms including Forbes, TechCrunch, Vellum, Finout and Artificial Analysis are cited as corroborating or discussing the economics.
The same material says Fable 5 and Mythos 5 use the same underlying model weights and carry identical pricing, differing in safeguards. It also says earlier Opus 4 and 4.1 models were priced higher at $15/$75, but are not presented as the active pricing benchmark for today’s lineup.
The key comparison in the source is not whether Fable 5 is stronger than Opus 4.8. It is whether the measured performance lift, the task-specific gains and the available discounts are enough to support a higher 2026 AI budget.
enterprise AI language models
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Proof Gap Around Productivity
It is not yet clear whether Fable 5’s higher price will translate into measurable productivity gains across typical enterprise work. The source material says there are zero controlled human-baseline productivity studies in the public record and one human-baseline enterprise proof point.
The Stripe migration account may be meaningful, but it remains a vendor-cited customer story rather than independently audited evidence. Details such as task scope, team composition, review burden, failure rate and comparable results from Opus 4.8 or Sonnet 4.6 are not established in the supplied material.
Another open question is workload fit. The benchmarks cited show stronger results in some areas and thinner gains in others, so buyers cannot assume the 5.7% aggregate lift maps cleanly to every coding, research, support or operations task.
AI model benchmark performance
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Procurement Tests Move Center Stage
The next step for buyers is likely to be workload-specific evaluation: testing Fable 5 against Opus 4.8 and lower-priced models on the exact tasks that drive spending. The supplied figures make clear that published pricing is settled, while productivity claims still need stronger public support.
More evidence may come from customer case studies, independent audits, benchmark updates and real deployment data after general availability. Until then, the decision to pay the Fable 5 premium rests on whether organizations can verify gains in their own systems, not on headline benchmark rank alone.
AI prompt caching discounts
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the confirmed news about Claude Fable 5?
Claude Fable 5 is described in the source material as generally available around July 1, 2026 and priced at $10 per million input tokens and $50 per million output tokens.
How much more expensive is Fable 5 than Opus 4.8?
The supplied pricing comparison says Fable 5 costs twice as much as Claude Opus 4.8, which is listed at $5/$25 for input and output tokens.
Does the benchmark data show Fable 5 is better?
The source cites Artificial Analysis scores showing Fable 5 ahead of Opus 4.8, including 64.9 versus 61.4 on the Intelligence Index. The size of the average gain is smaller than the price increase.
Is there public proof that Fable 5 makes workers more productive?
The source material says public human-baseline productivity evidence is limited to one unaudited customer story attributed to Stripe, with no controlled human-baseline productivity studies cited.
Should companies pay for Fable 5?
This article is not financial or procurement advice. Based on the source material, buyers would need to test their own workloads and compare results against cheaper models before treating the premium as justified.
Source: Thorsten Meyer AI
