🔍 Read the full analysis: Claude Fable 5.2 Cost Breakdown: Comparing Effort Levels By Task on ThorstenMeyerAI.com
Get hardware and tech essentials 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
Artificial Analysis’ November 10 comparison puts Claude Fable 5.2’s Intelligence Index score between 54 and 65 across five effort settings. The reported average cost per index task rises from $3.10 at Low to $26.54 at Max, with High scoring 62 for $9.08. These figures describe one benchmark, not the cost or performance every user will see.
Claude Fable 5.2 scored between 54 and 65 on the Artificial Analysis Intelligence Index across five effort settings, while the reported average cost per benchmark task ranged from $3.10 at Low to $26.54 at Max. The comparison, accessed November 10, shows that choosing a higher setting can raise cost and waiting time substantially, even when the score changes only modestly.
Artificial Analysis tested Fable 5.2 on an Intelligence Index made up of nine evaluations covering agentic coding, business documents, science and knowledge. Its reported “cost per task” is the average API cost for one task in that test. The results are Low: score 54, $3.10 and 6.5 minutes; Medium: 56, $3.80 and 8.3 minutes; High: 62, $9.08 and 17.9 minutes; Xhigh: 62, $14.83 and 26.4 minutes; and Max: 65, $26.54 and 46.4 minutes.
The strongest score came at Max effort, which ranked first among 74 models in the index at the time of the source’s reporting. High ranked third with a score of 62, while Xhigh ranked second with the same rounded score. Compared with High, Max cost almost three times as much per task for three additional index points. The figures are benchmark results, not a guarantee of performance on a company’s own workload.
Anthropic lists Fable 5.2 API prices at $8 per million input tokens and $40 per million output tokens across effort levels. The source attributes the differing benchmark bills to the amount of model output used for reasoning: Artificial Analysis recorded 34.9 million output tokens across its Low-effort index run and 275 million at Max. Medium is the default in Claude apps and Claude Code, according to the supplied report.
Claude Fable 5.2 cost by effort level
Artificial Analysis compared scores, average task costs and completion times across five effort settings. These are results from one benchmark, not guaranteed outcomes for every workload.
Five effort settings
Higher effort raised benchmark cost and time; the displayed scores are rounded.
| Effort | Score | Cost per task | Time |
|---|---|---|---|
| Low | 54 | $3.10 | 6.5 min |
| Medium | 56 | $3.80 | 8.3 min |
| High | 62 | $9.08 | 17.9 min |
| Xhigh | 62 | $14.83 | 26.4 min |
| Max | 65 | $26.54 | 46.4 min |
Where the score moves
The largest score increase was from Medium to High: six points for an additional $5.28 per task.
Score rose from 56 to 62; cost increased from $3.80 to $9.08.
Displayed score stayed at 62 while cost rose about 63%, to $14.83.
Max added three index points, with average task time reaching 46.4 minutes.
More effort is not always better
Results differed across evaluations, so the highest index score may not suit every task.
- Intelligence Index score: 62
- Average task cost: $9.08
- FrontierCode 2.0 score: 49.1%
- Intelligence Index score: 65
- Average task cost: $26.54
- FrontierCode 2.0 score: 48.8%
What businesses should weigh
Choose settings around the work and the people or systems waiting for results.
Track token use
Benchmark costs are averages; actual API bills depend on input and output token use.
Account for waiting
Longer completion times may affect workflows where a person is waiting for an answer.
Test real tasks
Measure accuracy, cost and time on your own workload before choosing an effort setting.
The Cost of More Model Effort
The comparison gives organizations a concrete trade-off to evaluate: higher effort may improve results on some tasks, but it can also increase token use, API cost and response time. In this index, the largest score jump is from Medium to High: six points for an additional $5.28 per task. Moving from High to Xhigh raises the reported cost by about 63%, while the displayed score remains 62. Max adds three points over High but takes an average of 46.4 minutes per task.
That difference can matter beyond an API invoice. A workflow with a person waiting for an answer may be affected by the longer completion time, while a batch process may be more tolerant of delay. The source also reports that Fable 5.2 at High scored 49.1% on Artificial Analysis’ FrontierCode 2.0 coding test, slightly above its 48.8% result at Max. More effort did not improve every reported result.
Artificial Analysis also reported that Fable 5.2 at High scored 62 on its index for $9.08 per task, compared with Claude Opus 5.5 at Max, which scored 59 for $10.04. That is a useful comparison within the same test, but it does not establish which model is cheaper or better for every task. Actual costs depend on usage and the work being done.
Products Worth Considering
LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz
Bose Lifestyle Ultra Compact Wireless Smart Home Speaker Bundle, Black
Logitech MX Master 4 Ergonomic Wireless Mouse with Haptic Feedback, Graphite
As an affiliate, we earn on qualifying purchases.
Fable and Mythos Release Details
Anthropic announced Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026, replacing their respective 5.1 versions, according to the source. It describes Mythos as the same underlying model as Fable, but with different safeguards and access: Fable is available to users generally, while Mythos is restricted to organizations accepted into Anthropic’s Cyber or Life Sciences Verification Programs.
Anthropic says Fable’s safeguards block some high-risk cybersecurity requests and most research biology tasks. The source says blocked queries receive a response from an older model, Opus 4.8 for cybersecurity and Opus 5 for biology. Anthropic also said early testing produced about 55% fewer blocks than Fable 5.1. These are company-reported details, not findings established by the Artificial Analysis benchmark.
Anthropic says the new models generate answers up to 30% faster than their 5.1 versions. It lists an optional fast mode at up to 2.5 times the token-generation speed, with prices of $16 per million input tokens and $80 per million output tokens. The source also relays Anthropic’s claims about prerelease achievements, including work on electronic devices and a large, agent-assisted mathematical proof effort; those company accounts were not independently verified in the supplied material.
“The models write their answers up to 30% faster than the 5.1 versions.”
— Anthropic, according to the supplied report
Limits of the Benchmark Results
The reported prices and scores apply to Artificial Analysis’ specific nine-evaluation index. The supplied material does not establish how well the results predict performance, cost or speed for a particular business workload, nor does it provide a broader range of repeated runs or uncertainty estimates for the effort-level comparisons. Scores are rounded, so equal displayed scores do not necessarily mean identical underlying results.
It is also unclear from the supplied material how often users will need High, Xhigh or Max effort, or how token use and latency will vary with task complexity. Anthropic’s reported release claims and prerelease examples are company statements; the material does not say they were independently verified. API bills will depend on actual input and output token use, and the benchmark averages should not be treated as fixed per-task prices.
Match Effort to Real Tasks
The next practical step for teams is to test the settings on their own work, recording accuracy, token use, completion time and cost rather than assuming the highest index score will deliver the best result for every task. Anthropic’s stated guidance is to begin with Opus 5.5 for most coding and knowledge work and use Fable when a task calls for it. The supplied source gives no further scheduled benchmark release or company milestone.
These reported costs concern model use, not a recommendation to buy a financial asset. API spending can vary with usage and workload, and the figures do not guarantee a particular outcome. The comparison is best read as a snapshot of one benchmark, with real-world performance still to be measured by users.
Key Questions
How much does a Fable 5.2 benchmark task cost?
Artificial Analysis reported average costs from $3.10 at Low effort to $26.54 at Max on its Intelligence Index. These are averages for tasks in that benchmark, not fixed prices for every user request.
Which effort setting had the highest index score?
Max effort scored 65 and ranked first among 74 models in the reported index results. Its average task cost was $26.54 and its average task time was 46.4 minutes.
Is Max always the best setting?
No. The benchmark shows higher scores at Max on the Intelligence Index, but Fable 5.2 at High scored slightly above Max on the reported FrontierCode 2.0 test. The best setting depends on the task and the cost and latency a user can accept.
What is the difference between Fable 5.2 and Mythos 5.2?
According to the supplied report, they share an underlying model but differ in safeguards and access. Fable is available to users generally; Mythos is restricted to organizations accepted into Anthropic’s Cyber or Life Sciences Verification Programs.
Are the benchmark costs what every customer will pay?
No. The figures are Artificial Analysis averages for its test, while actual API costs depend on token use and the work performed. Anthropic lists standard rates of $8 per million input tokens and $40 per million output tokens for Fable 5.2.
Source: ThorstenMeyerAI.com
