🔍 Read the full analysis: Is The AI Intelligence Gain Big Enough To Match The Price? on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Anthropic’s Claude Fable 5 costs $10 per million input tokens and $50 per million output tokens, twice the price of Opus 4.8. Artificial Analysis scores it 5.7% higher on its Intelligence Index, but the source review found no controlled human-baseline productivity study and only one unaudited customer account.
Anthropic’s Claude Fable 5 costs twice as much as Opus 4.8, while scoring 5.7% higher on Artificial Analysis’s Intelligence Index, according to a July review of model pricing and public evidence. The review found no controlled study measuring Fable 5’s productivity against a human-work baseline; its public enterprise case rests on one unaudited customer testimonial.
Anthropic lists Fable 5 at $10 per million input tokens and $50 per million output tokens. Opus 4.8 is priced at $5 and $25, respectively, while Sonnet 4.6 is listed at about $3 and $15. The review says the prices are corroborated by Anthropic’s launch materials and pricing documentation, as well as coverage and pricing references from other publications and services.
Artificial Analysis reports an Intelligence Index score of 64.9 for Fable 5, compared with 61.4 for Opus 4.8. On GDPval-AA, a benchmark for knowledge-work tasks, Fable 5 scores 1,932 Elo against Opus 4.8’s 1,890. The review characterizes that difference as a 2.2% increase, while noting that Elo ratings are not straightforward percentages of productivity.
The review estimates that a full Intelligence Index run costs about $9,940 for Fable 5 and $4,970 for Opus 4.8. Those are benchmark evaluation costs, not a forecast of what a company will spend in ordinary use. Actual bills depend on token volume, caching, output length and other pricing options; the analysis gives a blended estimate of $7.70 per million tokens under a specified cache-hit, input and output mix.
Pricing & evidence review · July 2026
Is The AI Intelligence Gain Big Enough To Match The Price?
Claude Fable 5 costs twice as much as Opus 4.8 and leads on an aggregate benchmark. The public evidence still does not show whether that lead translates into broadly higher workplace productivity.
What the premium buys
List prices show a clear cost gap. Caching and batch options can change the effective rate, while actual bills depend on token mix and workload.
Claude Fable 5
Claude Opus 4.8
Claude Sonnet 4.6
Artificial Analysis · Intelligence Index
Fable 5’s score is 5.7% higher on this aggregate index. This is a benchmark comparison, not a productivity percentage.
lead reported
Fable 5 scores 1,932 Elo against Opus 4.8 at 1,890. Elo ratings describe relative benchmark performance; they do not translate directly into worker productivity.
A promising story, limited proof
The analysis distinguishes a customer account from a controlled, independently reviewed productivity result.
The review reports no controlled study comparing Fable 5 productivity with a human-work baseline.
Stripe described a codebase-wide migration completed in one day, compared with more than two months by hand.
The review confirmed the testimonial appeared in Anthropic’s launch post, but found no audit or replication.
“A codebase-wide migration in a day that would otherwise have taken a whole team over two months by hand.” Stripe, as quoted in Anthropic’s launch post
The review says the public account does not establish whether scope, quality checks, review time or follow-up work were included. It also does not show that the result generalizes across teams or codebases.
Rates change with usage
Headline prices are only one part of an inference bill. The review’s blended estimate depends on a specific token mix and pricing assumptions.
Listed at $1 per million tokens, a 90% discount from standard input pricing.
$5 per million input tokens and $25 per million output tokens, according to the review.
About $7.70 per million tokens under a 7:2:1 mix of cache reads, regular input and output.
Estimated cost of a full Intelligence Index run for Fable 5 and Opus 4.8, respectively. These are evaluation costs, not forecasts of ordinary company usage.
Measure the work that matters
Aggregate benchmarks can identify a lead. A representative pilot can test whether that lead is worth paying for.
Choose real tasks
Use representative coding, research or knowledge-work tasks from your team.
Compare models
Run Fable 5 against Opus 4.8 or another lower-cost option under the same conditions.
Track full effort
Record token costs, completion, corrections, review time and delivered quality.
Judge the value
Compare end-to-end cost with measurable work delivered, task by task.
At a glance
How much does Fable 5 cost?
The July analysis lists $10 per million input tokens and $50 per million output tokens. Options such as caching, batch processing and U.S.-only inference can affect rates.
How does it compare with Opus 4.8?
Fable 5 costs twice as much at list price and scores 64.9 versus 61.4 on the Intelligence Index: a 5.7% benchmark difference, not a measured productivity gain.
Is its productivity gain proven?
The review reports no controlled human-baseline study. Its public enterprise example is one customer testimonial without an independent audit.
The Cost of a Small Average Lead
For organizations choosing a model for a large deployment, a twofold list-price difference can outweigh a modest average benchmark advantage if the more expensive model does not improve the work that matters to them. The review’s central point is that current public evidence does not establish how much Fable 5 improves employee output across typical business tasks, or whether that improvement offsets its cost.
That does not establish that the premium is poor value for every buyer. The analysis says Fable 5 set records in five of 10 sub-benchmarks, suggesting that performance can vary by task. A model that performs better on a company’s most demanding coding, research or knowledge-work tasks could be worth a higher rate, even if its overall score moves less. Buyers would need to test their own workloads and compare end-to-end costs rather than rely on a single aggregate score.
The distinction matters when setting budgets and making productivity claims. A benchmark measures performance on its own tasks and scoring rules; a customer story describes a particular deployment. Neither alone establishes a general productivity gain across an organization. The review argues that the price is documented more firmly than the productivity case, so companies should treat broad return-on-investment claims as unproven unless supported by their own measurements.
Products Worth Considering

Apple Mac mini (2018), 3.0 GHz 6-Core Intel Core i5, 32 GB RAM, 256 GB SSD

Logitech MX Ergo S Advanced Wireless Trackball Mouse USB-C – Graphite

Logitech MX Master 4 Ergonomic Wireless Mouse with Haptics – Graphite
As an affiliate, we earn on qualifying purchases.
How the Price Comparison Works
Fable 5 is the most expensive model in Anthropic’s active lineup, according to the review. It reports that prompt-cache reads receive a 90% discount, with a listed cache-read rate of $1 per million tokens, and that batch pricing is $5 per million input tokens and $25 per million output tokens. The blended $7.70 estimate assumes a 7:2:1 mix of cache reads, regular input and output tokens; it is an illustrative calculation, not a universal rate.
The source also says U.S.-only inference is available at a 1.1× pricing multiplier, and that Fable 5 and Mythos 5 use the same underlying weights but differ in safeguards. The review cautions that a comparison with GPT-5.5’s input price depends on a single secondary source, so it treats that comparison as directional rather than definitive.
Anthropic’s launch post, as described in the analysis, cites a Stripe account involving a 50-million-line Ruby codebase. Stripe said Fable 5 completed a codebase-wide migration in one day that would otherwise have taken a team more than two months by hand. The review verified that this statement was made, but says it found no accompanying methodology, independent audit or replication establishing the result as a controlled productivity measurement.
“A codebase-wide migration in a day that would otherwise have taken a whole team over two months by hand.”
— Stripe, as quoted in Anthropic’s launch post
What the Productivity Evidence Shows
The review reports zero controlled human-baseline productivity studies for Fable 5 and one enterprise proof point: the Stripe testimonial. It says stronger descriptions of the Stripe example as a concrete, independently established baseline were not supported by the material it checked. The testimonial remains a claim by a customer reported in Anthropic’s launch post, rather than a result independently audited in the review.
It is not clear whether the migration’s scope, quality checks, review time or follow-up work were included in the comparison with two months of manual work. The public material described in the analysis also does not establish whether the result generalizes to other codebases, industries or teams. Nor does the average benchmark score show which buyers will see meaningful gains on their own tasks.
The benchmark numbers themselves have limits: the 5.7% Intelligence Index difference is a comparison on an aggregate evaluation, not a percentage increase in worker productivity. The review notes that Fable 5’s leads vary across sub-benchmarks, but the supplied material does not provide enough detail to determine how those task-level results map onto a particular company’s work or cost savings.
What Buyers Should Measure
The next useful evidence would be controlled, independently reviewed evaluations that compare Fable 5 with lower-cost alternatives on representative enterprise tasks. Those studies would need to define what counts as completed work, account for human review and correction, and report both quality and time or cost—not just model benchmark scores.
Until such evidence is public, companies considering the premium can run a limited pilot against Opus 4.8 or another model on their own tasks. They can track token and inference costs, completion rates, error correction, review time and the value of work delivered. This would test whether Fable 5’s stronger results on particular benchmarks correspond to a measurable advantage in their own workflows.
The July analysis places general availability at around July 1 after a June 9 launch. The source material does not identify a scheduled independent productivity study or a further benchmark release. For now, the practical question for buyers remains whether their workload benefits enough from Fable 5’s measured lead to justify its higher rate.
Key Questions
How much does Claude Fable 5 cost?
According to the pricing figures in the July analysis, Anthropic charges $10 per million input tokens and $50 per million output tokens. Rates can differ with options such as prompt caching, batch processing and U.S.-only inference.
How does Fable 5 compare with Opus 4.8?
Fable 5 costs twice as much at list price. Artificial Analysis scores it 64.9 on the Intelligence Index, against 61.4 for Opus 4.8, a 5.7% difference on that benchmark—not a measured productivity increase.
Has Fable 5’s productivity gain been independently proven?
The source analysis reports no controlled human-baseline study for Fable 5. Its public enterprise example is a Stripe testimonial included in Anthropic’s launch post; the review says it found no independent audit or replication.
Could Fable 5 still be worth the higher price?
Possibly, for tasks where its performance advantage is useful. The analysis reports records in five of 10 sub-benchmarks, but buyers need to test their own work and compare quality, review time and total costs before drawing that conclusion.
Source: ThorstenMeyerAI.com
