🔍 Read the full analysis: Twice The Price Tag For A 5.7% AI Intelligence Increase 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 is priced at $10 per million input tokens and $50 per million output tokens, twice Opus 4.8’s rates. Artificial Analysis reports a 5.7% higher Intelligence Index score for Fable 5, but the cited source review found no controlled productivity study and one unaudited customer testimonial.
Anthropic’s Claude Fable 5 costs twice as much per token as Claude Opus 4.8 while scoring 5.7% higher on Artificial Analysis’s Intelligence Index, according to a review published by ThorstenMeyerAI.com. The review says public evidence that Fable 5 raises enterprise productivity is much thinner: it identifies no controlled human-baseline study and one unaudited customer account cited in Anthropic’s launch post.
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 review says the prices are documented in Anthropic’s launch materials and platform pricing information, and corroborated by several third-party publications and services. Fable 5 is the most expensive model in Anthropic’s active lineup, according to the review.
Artificial Analysis gives Fable 5 an Intelligence Index score of 64.9, versus 61.4 for Opus 4.8. The review calculates the difference as 5.7%. On GDPval-AA, a knowledge-work benchmark expressed as an Elo rating, Fable 5 scores 1,932 against Opus 4.8’s 1,890. These results describe benchmark performance; they do not establish how much more productive employees will be in a particular company.
The source puts the cost of a full Intelligence Index run at about $9,940 for Fable 5 and $4,970 for Opus 4.8. It also calculates a blended Fable 5 rate of $7.70 per million tokens using a stated mix of cached, input and output tokens. That estimate assumes seven cache reads, two input units and one output unit for every ten units, and is not the same as the list price for every workload.
Twice The Price Tag For A 5.7% AI Intelligence Increase
Claude Fable 5 costs twice as much per token as Opus 4.8. Its benchmark lead is measurable; a broad workplace productivity gain is not established by the cited public evidence.
Per million tokens
Per million tokens
Fable 5 / Opus 4.8
Unaudited customer account cited
What the premium costs
Published token rates make the price gap straightforward to compare. Actual bills depend on caching, batch use, geography and workload mix.
Standard rates
Anthropic’s listed rates, as reported in the review.
Standard rates
Half of Fable 5’s listed input and output rates.
Rates can vary
US-only inference is listed at a 1.1× multiplier: $11 input and $55 output per million.
A lead on the index, with limits
Benchmark scores describe results on selected tasks. They do not directly measure employee output or company return on investment.
Artificial Analysis Intelligence Index
Reported score comparison on a 0–100 scale.
1,932 vs 1,890 Elo
Fable 5’s reported score is 42 Elo points above Opus 4.8. Elo differences are not a simple percentage measure of head-to-head performance.
One customer story is not a controlled study
The review found a public example attributed to Stripe in Anthropic’s launch post, with no independent audit identified.
“In a 50-million-line Ruby codebase, Fable 5 performed a codebase-wide migration in a day that ‘would otherwise have taken a whole team over two months by hand.’”
Anthropic launch post, quoting a Stripe account
The review assesses the public material it examined. It cannot rule out private evaluations or undisclosed customer data.
Measure the work that matters to you
A benchmark advantage may be valuable on difficult, costly tasks. The useful answer depends on a buyer’s own workflow and total effort.
Representative tasks
Use real work that reflects your team’s workload and quality needs.
Same conditions
Run Fable 5 and Opus 4.8 with matching prompts and review standards.
Full effort and cost
Record completion time, quality, corrections, retries and token use.
Workload by workload
Check whether the measured gains offset the higher price for your use.
For companies choosing a model for large-scale use, the difference between a published price and a demonstrated business outcome is material. Token charges can be compared directly, while any productivity return depends on the task, the amount of human review, integration costs and how often the model succeeds without rework. The review’s central finding is that buyers can verify the twofold price gap more readily than they can verify a broad productivity claim.
The benchmark results do not mean Fable 5 is only 5.7% better for every task. The source says it leads in five of ten sub-benchmarks, with larger gains on some tasks; it also notes that a 42-point Elo difference should not be treated as a simple percentage measure of head-to-head performance. For a buyer, the practical question is whether the company’s own workload resembles the tasks where Fable 5 performs especially well. That calls for task-specific testing, rather than assuming the aggregate score predicts savings or output gains.
Conversely, the source review does not establish that the higher price is unjustified. A benchmark advantage on a costly or difficult task may be valuable even if the overall index gain is modest. The evidence presented supports a comparison of listed prices and benchmark scores; it does not provide a general return-on-investment calculation.
AI model performance benchmarking tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Pricing and Evidence Behind the Comparison
The article identifies Anthropic’s launch post, platform pricing documentation and Fable product page as sources for the listed rates. It says Fable 5 also has a 90% prompt-caching discount, with cache reads at $1 per million tokens, and a batch rate of $5 per million input tokens and $25 per million output tokens. For US-only inference, the source gives a 1.1-times price multiplier, or $11 and $55. Those options can affect a customer’s actual bill, depending on usage and requirements.
The review also says Fable 5 and Mythos 5 share the same underlying weights and are priced identically, with safeguards distinguishing the products. It reports the launch date as June 9, 2026 and general availability at around July 1. The comparison is specifically with Opus 4.8, not a claim that every competing model costs less or performs worse.
For productivity, the review traces the public case to one example in Anthropic’s launch post. Anthropic attributes to Stripe the account that Fable 5 completed a migration across a 50-million-line Ruby codebase in one day, a task said otherwise to have taken a team more than two months by hand. The source review says this survived verification as a statement made by Anthropic, but not as an independently audited measurement. Its findings should be read as an assessment of the cited public material, not proof that no additional private customer data exists.
“In a 50-million-line Ruby codebase, Fable 5 performed a codebase-wide migration in a day that “would otherwise have taken a whole team over two months by hand.””
— Anthropic’s launch post, quoting a Stripe account
What the Productivity Evidence Cannot Show
The material provided does not include a controlled study comparing Fable 5 with a human or no-AI baseline, an independent audit of the Stripe example, or a replicated result across other businesses. The methodology behind the reported migration, including how the task was scoped, what review was required and how the time comparison was calculated, is not detailed in the account summarized by the review.
It is also unclear how the benchmark scores translate to individual enterprise workloads, what the total cost would be after retries and human review, and whether the model’s strongest sub-benchmark results apply to a given buyer. The available information does not establish a general productivity percentage or return on investment. The review’s conclusion concerns the public evidence it examined; it cannot rule out private evaluations or results not disclosed in the cited sources.
How Buyers Can Test the Claims
The next useful evidence would be published evaluations that describe the tasks, comparison group, human oversight, error rates and full costs—not just model scores or a single customer example. Companies considering Fable 5 can compare it with Opus 4.8 on representative work using the same prompts and review standards, then track completion time, quality, correction effort and token use. Such tests would show whether the benchmark advantage offsets the higher price for their own work; results from one organization should not automatically be generalized to another.
Anthropic’s published pricing and benchmark comparisons remain the available basis in the source material. No further controlled productivity findings or independent replication are specified there. Until such evidence is released, broad claims about Fable 5’s workplace productivity should be treated as unproven by the cited public data, while its price and benchmark results can be assessed separately.
Key Questions
How much does Claude Fable 5 cost?
The source lists Anthropic’s rates as $10 per million input tokens and $50 per million output tokens. It also describes lower rates for cached and batch usage, plus a 1.1-times multiplier for US-only inference.
How does Fable 5 compare with Opus 4.8?
Fable 5’s listed input and output prices are twice Opus 4.8’s. Artificial Analysis reports an Intelligence Index score of 64.9 for Fable 5 and 61.4 for Opus 4.8, a 5.7% difference as calculated in the source review.
Does the 5.7% score gain mean 5.7% higher workplace productivity?
No. The figure refers to the Intelligence Index benchmark comparison, not a measured increase in employee output. The source review says it found no controlled Fable 5 productivity study against a human or no-AI baseline.
What is the evidence for Fable 5’s productivity claims?
The review identifies one public customer account in Anthropic’s launch post, attributed to Stripe, about a code migration in a large Ruby codebase. It says the claim was not accompanied by an independent audit or replication in the material it examined.
Is Fable 5 worth the higher price?
The cited evidence does not establish a universal answer. The benchmark advantage may matter more on some tasks than others, so buyers would need to test their own workloads and total costs rather than infer a return from the aggregate score alone.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
