🔍 Read the full analysis: AI For Twice The Price: Measuring A 5.7% Intelligence Boost on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

• Business-only prices and quantity discounts
• Tax-exempt purchasing
• Multiple users, one account, clear invoices

Create a free Business account →Free to register

As an affiliate, we earn on qualifying purchases.Hide

Anthropic’s Claude Fable 5 costs twice as much as Opus 4.8, while Artificial Analysis scores it 5.7% higher on its Intelligence Index, according to a report published by ThorstenMeyerAI.com. The report says the public case for a corresponding productivity gain is much thinner: it identifies one unaudited customer anecdote, not a controlled study comparing work output with a human-only baseline.

Anthropic lists Fable 5 at $10 per million input tokens and $50 per million output tokens. Opus 4.8 is listed at $5 and $25, respectively, making Fable 5’s rates twice as high. The report places Fable 5 at roughly 3.3 times the listed price of Sonnet 4.6, whose rates it gives as approximately $3 per million input tokens and $15 per million output tokens. Those figures concern token prices, not a buyer’s total cost, which varies with usage and workload.

For benchmark results, the report cites Artificial Analysis: Fable 5 scored 64.9 on the Intelligence Index, against 61.4 for Opus 4.8. On GDPval-AA, a knowledge-work benchmark, the scores were 1,932 and 1,890. The source describes that difference as 2.2%. It also estimates that a full Intelligence Index run costs about $9,940 for Fable 5, compared with about $4,970 for Opus 4.8. These are benchmark and evaluation costs, not a forecast of what an organization will spend in everyday use.

The productivity evidence is different in kind. Anthropic’s launch post recounts that Fable 5 completed a migration across a 50-million-line Ruby codebase in one day, a task said to have taken a team more than two months by hand. The report attributes the example to Stripe but says it found no published methodology, independent audit or replication for the account. It treats the story as evidence that a customer claim was made, not as independently measured proof of a general productivity effect.

The Case for Paying More

The figures give companies a clearer basis for comparing listed prices and benchmark scores than for predicting employee output. A 5.7% Intelligence Index advantage may matter to a buyer whose tasks align with areas where Fable 5 performs better; it does not by itself show that a team will finish work 5.7% faster or save enough money to cover the higher token bill.

The report says Fable 5 set records in five of ten sub-benchmarks, suggesting that an overall average can conceal larger gains on particular tasks. A company handling those tasks may reach a different purchasing decision from one whose work resembles the other benchmark categories. Buyers would need to test representative tasks and account for their own token usage, quality requirements and review costs. The report’s central point is that the price is documented more firmly than workplace productivity.

Comparing Models and Evidence

Fable 5 is described in the source as Anthropic’s most expensive active public model. The article dates its launch to June 9, 2026, and says general availability followed around July 1. Its prices are cited as confirmed by Anthropic’s launch post and pricing materials, with secondary coverage offered as corroboration. The report also notes that prompt caching and batch pricing can lower costs for some use cases; the headline rates are not the only available pricing structure.

The benchmark comparison answers a narrower question than an enterprise productivity study. An aggregated index summarizes performance across selected tests, while GDPval-AA reports an Elo rating for knowledge-work tasks. Neither directly measures the effect of deploying a model on a company’s staff, workflows or costs. The report also cautions against treating the 42-point GDPval-AA difference as a simple percentage description of head-to-head performance, because Elo ratings are logarithmic.

For the workplace claim, the report says its review of available evidence found one human-baseline example: the Stripe account in Anthropic’s launch materials. It distinguishes that customer story from controlled research, audited results and independent replication. That distinction matters because a striking project example can be relevant without establishing what other organizations should expect.

“would otherwise have taken a whole team over two months by hand”

— Anthropic launch post, as quoted in the ThorstenMeyerAI.com report

What the Productivity Claim Shows

The source does not provide a controlled study comparing Fable 5-assisted work with work completed by people without AI. It also does not identify an independent audit or replication of the Stripe account. Details such as how the migration was evaluated, what human review was required and whether the claimed timing includes the full project process are not established in the material provided.

The benchmark results also do not resolve whether the premium is worthwhile for a particular buyer. The source reports aggregate scores and sub-benchmark records, but does not establish how a company’s own tasks would perform or what its total costs would be. Claims that Fable 5 is more productive across enterprise work would go beyond the evidence described here.

Evidence Buyers Still Need

The next useful evidence would be independent, controlled evaluations of representative workplace tasks, with a clear human-only comparison, methods and results. Replication of the Stripe migration account, or additional customer cases with enough detail to assess their methods, could also help clarify how far the example generalizes.

Until such results are available, organizations weighing Fable 5 against lower-priced models can compare Anthropic’s published rates and the cited benchmark scores, then test their own workflows before committing to broad deployment. The report does not identify a scheduled study or a forthcoming independent productivity assessment, so when stronger evidence will appear remains unknown.

Key Questions

How much does Claude Fable 5 cost?

The report gives Anthropic’s listed rates as $10 per million input tokens and $50 per million output tokens. Actual costs depend on usage and may differ when caching or batch pricing applies.

What does the 5.7% figure measure?

It refers to Fable 5’s Intelligence Index score of 64.9, compared with 61.4 for Opus 4.8, according to Artificial Analysis as cited in the report. It is not a measured increase in employee productivity.

Is there independent proof Fable 5 makes workers more productive?

The report says it found no controlled human-baseline productivity study for Fable 5. It identifies a Stripe customer anecdote in Anthropic’s launch post, but says the account has not been independently audited or replicated in the material reviewed.

Does the benchmark result mean Fable 5 is worth twice the price?

Not on its own. The score is an aggregate benchmark result, and the report says performance varied across sub-benchmarks. Whether the extra cost pays off depends on a buyer’s tasks, usage and results in its own evaluation.

Source: ThorstenMeyerAI.com