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. Artificial Analysis benchmark data cited for the model shows a 64.9 Intelligence Index versus 61.4 for Opus 4.8, a 5.7% gain. The pricing is documented; broad productivity claims still rest mainly on an unaudited customer story.
Anthropic’s Claude Fable 5 has moved into public availability after its June 9 launch, putting its premium pricing under scrutiny: Anthropic lists the model at $10 per million input tokens and $50 per million output tokens, while Artificial Analysis data cited for the model shows a 5.7% Intelligence Index gain over Opus 4.8, which costs half as much.
The confirmed economics are straightforward. Fable 5 is priced at $10 input and $50 output per million tokens, compared with $5 and $25 for Opus 4.8 and roughly $3 and $15 for Sonnet 4.6. With a 7:2:1 cache-hit-to-input-to-output mix, the cited blended rate is $7.70 per million tokens.
The benchmark case is narrower. Artificial Analysis data cited in the review puts Fable 5 at 64.9 on its Intelligence Index, compared with 61.4 for Opus 4.8. On GDPval-AA, Fable 5 is cited at 1,932 Elo versus 1,890, a smaller aggregate lead than the price gap suggests.
The productivity evidence is thinner. Anthropic’s launch post credits Stripe with using Fable 5 on a 50-million-line Ruby codebase migration in one day, compared with work that it said would have taken a team more than two months by hand. That is an attributed customer account, not a controlled study, and the reviewed public record cites no human-baseline enterprise trial.
AI Budgets Face New Math
For buyers, the issue is not whether Fable 5 is stronger; the question is where that strength pays for itself. A model that doubles token cost may be worth it for high-value coding migrations, long-running agents, or research tasks, but broad rollout needs proof that task savings exceed inference spend.
The risk is budget drift. If teams use Fable 5 where Opus 4.8 or a Sonnet-class model is enough, costs can rise before managers see measurable gains. The cited full Intelligence Index run cost of about $9,940 versus about $4,970 for Opus 4.8 captures the trade-off.
AI language model API pricing
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Launch to Availability
Axios reported on June 9 that Anthropic released Fable 5 as a general-use Mythos-class model with safeguards. Tom’s Hardware reported the $10/$50 API price and the comparison with Opus 4.8 and Sonnet 4.6.
The Guardian reported on July 1 that customer access had been restored after a more than two-week blackout tied to U.S. export controls and security concerns. That timing makes the price-performance debate immediate for enterprise customers deciding whether to resume or expand use.
“would otherwise have taken a whole team over two months by hand”
— Anthropic launch post, citing Stripe
enterprise AI model comparison
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Productivity Proof Still Thin
It is not yet clear whether Fable 5’s benchmark lead translates into repeatable workplace gains across finance, coding, research, or operations. The public record reviewed here includes zero controlled human-baseline studies and one unaudited customer story, so broad productivity claims remain vendor-attributed.
It is also unclear how often prompt caching, batch pricing, and task routing will lower real customer costs. The blended $7.70 per million tokens rate depends on a cache-heavy usage pattern that not every workload will match.
AI model benchmark tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Buyer Pilots Meet Benchmarks
The next test is whether customers can show controlled before-and-after results for the tasks that justify top-line model spend. Procurement teams are likely to compare Fable 5 with Opus 4.8 and Sonnet-class models on their own data before committing larger budgets.
Watch for audited case studies, workload-specific benchmark releases, and updated pricing from Anthropic or rivals. For now, the confirmed story remains narrow: Fable 5 costs more, scores higher on aggregate benchmarks, and lacks public controlled productivity proof at the same level.
AI coding migration tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
How much does Claude Fable 5 cost?
Claude Fable 5 is listed at $10 per million input tokens and $50 per million output tokens, twice the cited price of Opus 4.8.
Is Fable 5 better than Opus 4.8?
On the cited aggregate benchmark, yes. Artificial Analysis data puts Fable 5 at 64.9 versus 61.4 for Opus 4.8, a 5.7% gain.
Does the 5.7% gain mean Fable 5 is not worth buying?
Not necessarily. Average benchmark gains may hide larger gains on specific tasks, so the value depends on workload fit, cache usage, and whether the model reduces labor enough to offset price.
What productivity proof has Anthropic shown?
The main public proof point is Stripe’s code migration story, attributed by Anthropic. It is a customer account, not an independent controlled study.
What should enterprise buyers do next?
Buyers should run side-by-side pilots against Opus 4.8 and cheaper models, measure task completion quality, and calculate actual token costs under their own workload patterns.
Source: Thorsten Meyer AI
