🔍 Read the full analysis: A Guide To Claude Fable 5.2 Effort Levels And Per-Task Costs 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
Anthropic released Claude Fable 5.2 on November 9, 2026, with five effort settings that trade test performance for cost and time. Artificial Analysis reports costs of $3.10 per task at Low and $26.54 at Max; its test score rose from 54 to 65. The figures describe one benchmark, not a guaranteed cost or result for every user.
Anthropic released Claude Fable 5.2 on November 9, and benchmark data published by Artificial Analysis shows how its five effort levels affect test scores, time and estimated cost. Across the index, the reported cost ranges from $3.10 per task at Low to $26.54 at Max, while the score rises from 54 to 65—information that gives businesses a way to weigh added reasoning against a higher bill.
Artificial Analysis tested Fable 5.2 on its Intelligence Index, which covers nine evaluations in areas including agentic coding, business documents, science and knowledge. The index score is 54 at Low, 56 at Medium, 62 at High, 62 at Xhigh and 65 at Max. The reported average cost per task follows a steeper curve at the higher settings: $3.10, $3.80, $9.08, $14.83 and $26.54, respectively.
The same test reports average task times of 6.5 minutes at Low, 8.3 at Medium, 17.9 at High, 26.4 at Xhigh and 46.4 at Max. Artificial Analysis says each level has the same listed token rates—$8 per million input tokens and $40 per million output tokens—while the amount of model output used for the full index increases from 34.9 million tokens at Low to 275 million at Max. These per-task costs are benchmark estimates, not a flat fee for every real-world request.
The results show diminishing returns at the top of this particular test. High scores 62 for $9.08 per task; Xhigh also scores 62 as displayed, at $14.83. Max reaches 65, but costs almost three times as much as High and takes more than twice as long on average. Artificial Analysis also reports a separate FrontierCode 2.0 result in which Fable at High scored 49.1%, slightly above its 48.8% result at Max. Outcomes can vary by task and evaluation.
Effort levels trade cost for score
Artificial Analysis compared five effort settings on its Intelligence Index. Higher effort raised the top score, but also increased estimated cost and task time.
Five effort levels
Index scores, estimated task costs and average times rise across the settings, though gains are not even at every step.
| Effort | Score | Cost per task | Average 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 extra effort pays off
The displayed score is unchanged from High to Xhigh, while the estimated cost and time increase.
- Score: 62
- Estimated cost: $9.08
- Average time: 17.9 minutes
- Score: 62
- Estimated cost: $14.83
- Average time: 26.4 minutes
Choose settings for the work
Teams can compare quality, completion time and token use on their own tasks.
Use prompts and success criteria that reflect your real workload.
Track results, latency and token-based cost for each setting.
Use higher effort when its improvement justifies extra time and expense.
The Cost of More Model Reasoning
For organizations using AI for coding, research or document work, the effort setting can affect both the bill and how long a person waits. A higher setting may be useful when a task benefits from more extensive reasoning, but the index figures do not show a consistent payoff for every step up the ladder. In this benchmark, the largest displayed score increase is from Medium to High; Xhigh has the same rounded score as High at a higher estimated cost.
The practical lesson is to match effort to the work rather than automatically selecting Max. A business could test settings on its own workloads and track accuracy, completion time and token use. That is a conclusion drawn from the benchmark pattern, not a recommendation or performance guarantee from Artificial Analysis.
Products Worth Considering
Apple MacBook Pro 16-inch (2021), M1 Pro, 10-core CPU, 16-core GPU, 16GB RAM, 512GB SSD, Silver
LG UltraGear 27G640A-B 27-Inch QHD IPS Gaming Monitor, 300Hz, 1ms, G-SYNC Compatible, FreeSync Premium, DisplayHDR 400
LG UltraGear 32GX870A-B 32-inch 4K OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Fable, Mythos and Anthropic’s Pricing
Anthropic released Fable 5.2 and Mythos 5.2 as replacements for their 5.1 versions. According to the source material, the models share an underlying system, but their access and safeguards differ: Fable is available to users with protections for certain high-risk cyber and biology requests, while Mythos is offered to organizations accepted into Anthropic’s Cyber or Life Sciences Verification Programs. The reported effort-level benchmark concerns Fable.
Anthropic says the new models produce answers up to 30% faster than the 5.1 versions and lists standard pricing at $8 per million input tokens and $40 per million output tokens. It also offers an optional fast mode, with listed rates of $16 and $80 per million tokens. Anthropic recommends starting with Claude Opus 5.5 for most coding and knowledge work and moving to Fable when a task calls for it. Separately, Anthropic has described demanding prerelease demonstrations, including a multi-agent attempt to formalize a proof of Fermat’s Last Theorem; those accounts are vendor claims, not independent verification of everyday performance.
“The models write answers up to 30% faster than the 5.1 versions.”
— Anthropic, on the 5.2 models
Limits of the Benchmark Figures
The reported per-task costs come from Artificial Analysis’s Intelligence Index and should not be read as the price of any particular customer job. Actual use depends on the task, including how many input and output tokens it consumes, and the source does not give a universal cost for workplace use. The index scores are shown as rounded values, so equal displayed scores do not establish that two settings perform identically in every respect.
The source material does not provide full details here about the index’s cost calculation, uncertainty ranges or how results may change across repeated runs. Anthropic’s performance and safety claims—including its faster-generation comparison and reports of fewer blocked requests—are company statements. The dramatic prerelease demonstrations are also Anthropic accounts and should not be treated as independently verified typical outcomes.
Testing Settings on Real Workloads
Readers comparing Fable settings can use the benchmark as a starting point, then test relevant tasks with their own prompts and success criteria. The useful next comparison is not just a score: teams can record quality, latency and token-based cost at each effort level, and check whether a higher setting improves the result enough to justify the extra time and expense.
Further benchmark updates or independent tests could clarify how consistently the effort-level differences hold across tasks. Until then, the published figures describe Artificial Analysis’s evaluation as accessed November 10, 2026; they do not establish which setting is best for every user or workload.
Key Questions
How much does a Fable 5.2 task cost at each effort level?
Artificial Analysis reports estimated index costs of $3.10 at Low, $3.80 at Medium, $9.08 at High, $14.83 at Xhigh and $26.54 at Max. These are benchmark averages, not fixed prices for every real-world task.
Which effort level had the highest Intelligence Index score?
Max scored 65, the highest result reported for Fable 5.2 on this index. It also had the highest estimated cost and longest average task time in the five-level comparison.
Is Xhigh better than High?
On the displayed Intelligence Index results, both scored 62. Xhigh’s estimated task cost was $14.83, compared with $9.08 at High. The rounded scores do not establish that the settings behave identically on every task.
Are the costs the same for every Fable 5.2 request?
No. The figures are Artificial Analysis estimates for tasks in its index. Anthropic lists token prices, but the bill for an individual request depends on its token use and other applicable settings, including whether optional fast mode is used.
What is the difference between Fable 5.2 and Mythos 5.2?
According to the source material, they share an underlying model but differ in access and safeguards. Fable is available with protections for certain high-risk requests; Mythos is restricted to organizations accepted into Anthropic’s Cyber or Life Sciences Verification Programs.
Source: ThorstenMeyerAI.com
