🔍 Read the full analysis: A Closer Look At AI's 5.7% Gain And Double Price on ThorstenMeyerAI.com
Get audio and creator gear delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Anthropic’s Claude Fable 5 is priced at $10 per million input tokens and $50 per million output tokens, twice the rates for Opus 4.8. Artificial Analysis reports a 5.7% higher Intelligence Index score, while the public evidence for enterprise productivity gains consists of one unaudited customer account, not a controlled study.
Anthropic’s Claude Fable 5 costs twice as much as Opus 4.8 at list price, while Artificial Analysis reports a 5.7% higher Intelligence Index score for the newer model. A review published by ThorstenMeyerAI.com says the public case for greater enterprise productivity remains much thinner: it identifies one unaudited customer account and no controlled human-baseline productivity study.
Anthropic lists Fable 5 at $10 per million input tokens and $50 per million output tokens, compared with $5 and $25 for Opus 4.8. The source says this makes Fable 5 the most expensive model in Anthropic’s active lineup. Its listed rates are also about 3.3 times Sonnet 4.6’s approximate $3 input and $15 output prices.
On Artificial Analysis’s Intelligence Index, Fable 5 scored 64.9, against 61.4 for Opus 4.8, a reported 5.7% difference. On the knowledge-work benchmark GDPval-AA, Fable 5 scored 1,932 Elo to Opus 4.8’s 1,890, a 42-point gap. The source cautions that the Elo difference should not be reduced to a simple percentage interpretation: it describes the rating as logarithmic and says the result corresponds to roughly a 56% head-to-head win rate.
The source’s productivity evidence is a customer story included in Anthropic’s launch post. Anthropic said that, in a 50-million-line Ruby codebase, Fable 5 completed a codebase-wide migration in a day that would otherwise have taken a team more than two months by hand, attributing the account to Stripe. ThorstenMeyerAI.com says the story has no published methodology, audit or independent replication. It therefore establishes that the account was made, not that the result has been independently measured.
Model economics · Evidence review
A Closer Look at AI’s 5.7% Gain and Double Price
Claude Fable 5 posts higher benchmark scores at twice Opus 4.8’s list price. The public case for broad workplace productivity gains is still a single unaudited customer account.
$10 input and $50 output per million tokens
64.9 for Fable 5 versus 61.4 for Opus 4.8
Per million tokens
Per million tokens
1,932 vs. 1,890
No controlled study identified
01 / The premium
Paying twice as much changes the hurdle
At list price, Fable 5 is the most expensive model in Anthropic’s active lineup. Its rates are also about 3.3 times Sonnet 4.6’s approximate pricing.
Claude Fable 5
Anthropic list rates
Claude Opus 4.8
Half of Fable 5’s listed rates
Claude Sonnet 4.6
Approximate rates cited in the review
List price is a baseline, not every customer’s bill. Prompt caching can reduce cache reads to $1 per million tokens; batch rates are $5 input and $25 output. A modeled $7.70 blended rate assumes a specific 7:2:1 mix of cache hits, input and output.
02 / Benchmark results
A measurable lead, with a narrow meaning
Artificial Analysis reports higher scores for Fable 5 on two named evaluations. These results describe benchmark performance, not a forecast of employee output.
Artificial Analysis Intelligence Index
Composite score · reported difference: 5.7%
+3.5 INDEX POINTSGDPval-AA knowledge-work benchmark
Elo rating · roughly 56% head-to-head win rate
+42 ELO · NOT A SIMPLE % GAIN03 / What the evidence supports
A customer account is not a productivity study
Anthropic’s launch post attributes a striking migration story to Stripe. The source review says the account has not been independently audited or replicated.
A day instead of more than two months
In a 50-million-line Ruby codebase, Fable 5 reportedly completed a codebase-wide migration in one day that would otherwise have taken a team more than two months by hand.
The account exists; the result is unverified
The review identifies no published methodology, audit, or independent replication. Task conditions, quality checks, staffing assumptions, and verification are not provided in the summarized account.
04 / Buyer’s decision path
Make the premium prove itself on your work
The practical question is whether stronger results on relevant tasks offset higher inference costs. Compare the models under the same conditions before making a broad business case.
Choose representative tasks
Use real work and realistic difficulty.
Compare on equal terms
Run Fable 5 and Opus 4.8 against the same criteria.
Score quality and effort
Track accuracy, review time, speed, and token use.
Calculate total cost
Include caching, batch eligibility, and usage mix.
Timing & caveats
What the supplied review says
Fable 5 launched on June 9, 2026, with general availability expected around July 1, 2026. The review says Fable 5 and Mythos 5 share underlying weights and list pricing, with differences in safeguards.
It estimates a full Intelligence Index run at about $9,940 for Fable 5 and $4,970 for Opus 4.8. Those are benchmark-run costs under the evaluation’s assumptions, not typical customer bills.
The review’s GPT-5.5 price comparison relies on one secondary source, so it is directional rather than confirmed on the same footing as Anthropic’s listed rates.
Quick answers
Key questions
How much does Claude Fable 5 cost?
$10 per million input tokens and $50 per million output tokens at list price. Discounts and token mix can change actual costs.
What does the 5.7% figure mean?
It is the reported difference on Artificial Analysis’s Intelligence Index: 64.9 for Fable 5 versus 61.4 for Opus 4.8. It does not measure employee productivity.
Does the Stripe example prove productivity gains?
No. It is an attributed customer account without published methodology, audit, or independent replication in the supplied review.
Will it always cost twice as much in practice?
Not necessarily. Listed input and output rates are twice Opus 4.8’s, while caching and batch discounts can affect the bill.
The numbers pose a practical purchasing question: whether Fable 5’s higher scores justify paying twice Opus 4.8’s list price depends on the buyer’s tasks, token use and required performance. An average benchmark gain does not establish that every business workflow will improve by the same amount, nor that the improvement will offset higher inference costs.
The source notes that Fable 5 set records in five of ten sub-benchmarks, with larger gains on some tasks than the blended Intelligence Index suggests. That makes task-level results relevant to buyers. A model that performs better on a particular difficult workflow may justify a premium for that use, even if an overall composite shows a smaller gap. But without controlled studies of workplace outcomes, companies cannot use the cited testimonial alone as a dependable estimate of productivity or return on investment.
For budget planning, the distinction is between documented pricing and benchmark results and unverified claims about staff productivity. The first can be compared directly. The second needs evidence tied to actual users, tasks, time saved and costs before it can support a broad business case.
As an affiliate, we earn on qualifying purchases.
Pricing and Benchmark Baselines
The supplied analysis says Fable 5 launched on June 9, 2026, with general availability expected around July 1. It describes the prices as confirmed against Anthropic’s launch post, platform pricing documentation and product page, with additional reporting from other outlets and services. Those list prices are the clearest basis for comparison in the material.
Actual spending can vary with usage. According to the source, Anthropic offers a 90% prompt-caching discount, with cache reads priced at $1 per million tokens, and a batch rate of $5 per million input tokens and $25 per million output tokens. It calculates a blended $7.70 per million tokens using a specific 7:2:1 ratio of cache hits, input and output. That is a modeled rate, not a universal price: a customer’s bill depends on its own token mix and eligibility for discounts.
The source also says Fable 5 and Mythos 5 use the same underlying weights and have identical pricing, with differences in safeguards. It reports that Artificial Analysis’s full Intelligence Index run costs about $9,940 for Fable 5 and $4,970 for Opus 4.8. That benchmark-run cost is separate from a typical customer’s bill and reflects the evaluation’s workload and pricing assumptions.
Productivity Evidence Still Limited
The supplied material does not identify a controlled study comparing Fable 5 users with a human or no-AI baseline. It describes the Stripe account as the sole public enterprise proof point in that category and says it is unaudited. The methodology, task conditions, quality checks, staffing assumptions and independent verification are not provided in the account summarized here.
It is also unclear how broadly the benchmark results translate to business use. The source gives aggregate scores and says some sub-benchmarks show larger gains, but the material provided does not detail which customer workflows would see those gains or the costs of achieving them in production. Buyers should treat the 5.7% figure as a difference on a named benchmark index, not a forecast of a 5.7% increase in employee output.
The source’s comparisons with GPT-5.5 are less firmly supported: it says Fable 5’s input price is roughly twice as high, but notes that this comparison rests on one secondary source. That price comparison should be treated as directional rather than established on the same footing as the Anthropic list prices.
Verify Results Against Real Work
The next useful evidence would be independent, controlled productivity studies that disclose task selection, comparison groups, output quality, time saved and total costs. Until such results are public, the Stripe account remains an attributed customer anecdote rather than a general measure of Fable 5’s business impact.
For organizations considering the model, the practical next step is to test it on their own representative tasks and compare it with Opus 4.8 using the same evaluation criteria. That should include output accuracy and review effort, not only speed, as well as token consumption and the availability of caching or batch pricing. The source does not provide a date for additional independent studies or confirm that Anthropic will publish one.
Anthropic’s general-availability timing was listed as approximately July 1, 2026, following the June 9 launch. Whether the rollout brings further pricing details, customer data or audited productivity results is not established in the supplied material.
Key Questions
How much does Claude Fable 5 cost?
According to Anthropic pricing information cited by ThorstenMeyerAI.com, Fable 5 costs $10 per million input tokens and $50 per million output tokens at list price. Discounts and a customer’s input-output mix can affect actual costs.
How much higher is Fable 5’s Intelligence Index score?
Artificial Analysis reports a score of 64.9 for Fable 5, compared with 61.4 for Opus 4.8. The source describes the difference as 5.7% on the Intelligence Index; it is not a direct measurement of employee productivity.
Does the Stripe example prove Fable 5 improves productivity?
No. Anthropic’s launch post attributes a migration account to Stripe, but the source review says it has no published methodology, audit or independent replication. It is a customer testimonial, not a controlled productivity study.
Is Fable 5 always twice as expensive in practice?
No. Its listed input and output rates are twice Opus 4.8’s, but the supplied analysis describes prompt-caching and batch discounts. Actual costs depend on usage patterns and which pricing options apply.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
