🔍 Read the full analysis: Twice As Much For AI, With 5.7% More Intelligence on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get privacy and security gear delivered free with Prime

• Fast, free delivery on millions of items
• Prime Video, Amazon Music and more included
• Member-only deals all year

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

Anthropic’s Claude Fable 5 costs twice as much per token as Opus 4.8, while scoring 5.7% higher on Artificial Analysis’s Intelligence Index, according to a July 2026 analysis of the model’s price and public evidence. The analysis says the price is documented by Anthropic’s published materials, but enterprise productivity claims rest on one unaudited customer account, not controlled studies.

Anthropic lists Fable 5 at $10 per million input tokens and $50 per million output tokens. Opus 4.8 costs $5 and $25, respectively, while Sonnet 4.6 is listed at approximately $3 and $15. Those rates make Fable 5 the most expensive model in Anthropic’s active lineup, according to the analysis. It says the pricing appears in Anthropic’s launch post and product pricing documentation, and is also reported by several technology and industry publications.

Artificial Analysis’s reported Intelligence Index scores are 64.9 for Fable 5 and 61.4 for Opus 4.8. On GDPval-AA, a benchmark for knowledge-work tasks, the scores are 1,932 and 1,890. The analysis characterizes those results as a 5.7% Intelligence Index advantage and a 2.2% GDPval-AA difference. These are benchmark comparisons, not direct measurements of workplace productivity or a guarantee that a buyer will see equivalent gains.

The analysis also reports that Fable 5 sets records in five of ten sub-benchmarks, with gains varying by task. Its estimate of about $7.70 per million tokens is based on a stated mix of cached, input and output tokens; it is not the standard list price. The source says prompt-cache reads and batch processing are discounted, and US-only inference carries a 1.1-times price multiplier. Actual costs depend on usage patterns and configuration.

The Premium Depends on the Workload

For organizations choosing a model for large-scale use, token prices compound with volume. A model that costs twice as much as its predecessor needs to deliver enough value on a buyer’s particular tasks to justify that spending. The index gap alone cannot establish whether it will do so: benchmark scores summarize defined tests, while companies’ workloads, review requirements and costs differ.

The evidence gap matters because productivity claims can influence AI budgets. The source finds no controlled human-baseline enterprise study for Fable 5. It identifies one customer story about a code migration, but says the account lacks published methodology, an audit and independent replication. Buyers should treat that as a reported example, not a measured productivity rate that can be applied across teams.

There may still be a case for paying more where a model’s stronger results match high-value, difficult tasks. The analysis says Fable 5 leads on five sub-benchmarks, although the supplied material does not provide enough task-level detail to establish which customers would benefit or how large their savings might be. The practical question is what it costs and performs on a buyer’s own work, not just its overall ranking.

Pricing, Scores and the Launch Claim

Fable 5 launched on June 9, 2026, with general availability reported around July 1. The source describes it as Anthropic’s most powerful publicly available model at the time. Its comparison focuses on Opus 4.8, priced at exactly half Fable 5’s listed rates, and Sonnet 4.6, which costs less still.

The analysis distinguishes the published price from a blended-cost estimate. It calculates $7.70 per million tokens using a 7:2:1 ratio of cache-hit, ordinary input and output tokens, with cache reads priced at $1 per million. A different workload mix would produce a different effective rate. It also says Fable 5 and Mythos 5 use the same underlying weights and have identical prices, while differing in safeguards.

Anthropic’s launch post, as quoted in the source, attributes a codebase-wide migration in a 50-million-line Ruby codebase to Fable 5. The company’s post says the work was completed in a day and “would otherwise have taken a whole team over two months by hand,” crediting Stripe. The source’s verification accepts that this statement was made, but not that it is an independently established measurement. No methodology or independent replication is presented in the supplied material.

Workplace Gains Remain Unmeasured

The available material does not establish how Fable 5 changes productivity across an enterprise, whether the Stripe example can be independently reproduced, or how its reported benchmark gains translate into saved time or money. The source reports zero controlled human-baseline productivity studies and one unaudited customer account. It does not provide a verified productivity percentage, a representative set of company deployments or a comparison of total operating costs.

It is also unclear from the supplied material how the benchmark results map to specific customer workloads, how much the model’s performance varies across tasks, and whether the listed prices or benchmark scores have changed since the analysis was published. The five sub-benchmarks where Fable 5 reportedly sets records are not described in enough detail here to judge the relevance or size of those leads for a particular organization.

Test Costs Against Real Tasks

Organizations considering Fable 5 can compare it with Opus 4.8 and lower-cost alternatives on their own representative tasks, tracking output quality, human review time, token use and total cost. Such tests can help determine whether the higher price is justified for a specific workflow, but they would be local evaluations rather than proof of broad enterprise productivity gains.

Further evidence would require transparent study methods, a defined human or no-AI baseline, and results that can be independently checked. The supplied source does not identify a forthcoming controlled study or a scheduled update from Anthropic. Until such evidence is published, the pricing and benchmark comparisons are available, while the broader productivity case remains unverified.

Key Questions

How much does Claude Fable 5 cost?

The supplied July 2026 analysis gives Anthropic’s list prices as $10 per million input tokens and $50 per million output tokens. Discounts and actual spending depend on usage and configuration.

How does Fable 5 compare with Opus 4.8?

Fable 5’s listed input and output rates 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 characterized in the source.

Does the benchmark prove Fable 5 makes workers 5.7% more productive?

No. The 5.7% figure refers to the Intelligence Index comparison, not measured worker productivity. The source reports no controlled human-baseline enterprise productivity studies for Fable 5.

What productivity evidence does the source identify?

It identifies one account, attributed to Stripe in Anthropic’s launch post, about completing a code migration in a day. The analysis says the account has no published methodology, audit or independent replication, so it should not be treated as a general productivity measurement.

Source: ThorstenMeyerAI.com