🔍 Read the full analysis: Understanding Per-Task Costs For Every Claude Fable 5.2 Effort Level on ThorstenMeyerAI.com
Get the latest gadgets 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 reports that Claude Fable 5.2’s estimated cost per Intelligence Index task ranges from $3.10 at Low effort to $26.54 at Max, with scores rising from 54 to 65. High effort scored 62 for $9.08, while Max added three points at nearly three times the cost. These are benchmark estimates, not a guaranteed bill for every real-world task.
Claude Fable 5.2 costs an estimated $3.10 to $26.54 per task across five effort settings in Artificial Analysis testing, while its Intelligence Index score ranges from 54 to 65. The results, accessed November 10, show the trade-off for businesses choosing how much computation to buy: High effort scored 62 for $9.08, while Max scored 65 for almost three times the cost.
Artificial Analysis tested Low, Medium, High, Xhigh and Max effort on its Intelligence Index, which covers nine evaluations in areas including agentic coding, business documents, science and knowledge. Its per-task figure is the average API cost for one task in that test, not a fixed price for any individual customer workload. The published results list scores of 54 at Low, 56 at Medium, 62 at High, 62 at Xhigh and 65 at Max.
Estimated cost and average time per task rose with effort: $3.10 and 6.5 minutes at Low; $3.80 and 8.3 minutes at Medium; $9.08 and 17.9 minutes at High; $14.83 and 26.4 minutes at Xhigh; and $26.54 and 46.4 minutes at Max. Artificial Analysis ranks the Max result first among 74 models in the index. The source says Fable also placed third at High and second at Xhigh, based on the displayed rankings and rounded scores.
The source reports the same list prices across all effort settings: $8 per million input tokens and $40 per million output tokens. The difference in estimated task costs reflects how many tokens the model uses. Across the full index, reported output-token use rose from 34.9 million at Low to 275 million at Max. Those totals describe the test run, not a per-task token allowance.
Claude Fable 5.2 costs rise with effort
Artificial Analysis estimates higher benchmark scores at higher effort settings, alongside greater cost and longer task times. These are test averages, not guaranteed costs for real workloads.
Cost and score by effort
Each figure is an estimated average for one Intelligence Index task.
| Effort | Score | Cost | 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 |
What the trade-off looks like
The score rose by six points for $5.28 more per task.
Estimated cost rose by $5.75 while the displayed score stayed at 62.
Max added three score points at nearly three times the cost and more than twice the runtime.
Same token rates, more usage
The source reports identical list prices at every effort setting; estimated task costs rise as token use increases.
- $8 per million input tokens
- $40 per million output tokens
- 34.9 million tokens at Low
- 275 million tokens at Max
Use the benchmark as a guide
The index covers nine evaluations, including coding, business documents, science and knowledge.
May suit harder tasks
Additional benchmark performance may matter when the work is difficult and the gain is valuable.
Watch the added cost
Higher effort may be wasteful for routine jobs or tasks that need quick responses.
Test your own tasks
Compare quality, runtime and total API cost using representative prompts.
The Cost of More Model Effort
The results give teams a way to weigh benchmark performance against API spend and response time. In this test, the step from Medium to High added six score points for $5.28 more per task. Moving from High to Xhigh cost another $5.75, while the displayed score stayed at 62. Max added three points over High, but its estimated task cost was nearly three times as high and its average runtime was more than twice as long.
That trade-off matters when a company runs many tasks or needs quick responses. A high-scoring configuration may make sense for difficult work where additional performance is valuable; it may be wasteful for routine jobs. The benchmark does not establish which setting is best for a particular organization, and its score differences should not be treated as a promise of equal gains on other tasks.
The supplied source also reports that Fable 5.2 at High scored 49.1% on Artificial Analysis’s FrontierCode 2.0 test, slightly above its 48.8% result at Max. That is a reminder that higher effort did not improve every reported result. The source recommends comparing settings on the actual work a team plans to run.
Products Worth Considering
Samsung 32-Inch Odyssey G5 QHD 180Hz Gaming Monitor
LG UltraGear 32-Inch QHD 180Hz Curved Gaming Monitor
Samsung Odyssey G3 G30D 27-Inch FHD Gaming Monitor
As an affiliate, we earn on qualifying purchases.
How Fable 5.2 Is Positioned
According to the supplied report, Anthropic released Fable 5.2 on November 9, 2026, replacing Fable 5.1. It describes a separate model, Mythos 5.2, as the same underlying model with fewer safety restrictions, offered only to organizations accepted into Anthropic’s Cyber or Life Sciences Verification Programs. These release and access details come from the source material and are not independently verified here.
The report says Fable has safeguards that block some high-risk cybersecurity requests and most research-biology tasks; blocked requests are handled by older models. Anthropic reportedly said early testing showed about 55% fewer blocks than with Fable 5.1. Anthropic also lists prerelease achievements, including work on electronics and mathematical proofs. Those are vendor accounts, not independent evidence of typical customer performance.
Anthropic’s reported guidance is to start with Claude Opus 5.5 for most coding and knowledge work and move to Fable when a task requires it. In the Artificial Analysis comparison cited by the source, Opus 5.5 at Max scored 59 at an estimated $10.04 per task, while Fable 5.2 at High scored 62 at $9.08. The source says Opus 5.5 has lower token prices, making task-level cost and performance more informative than token rates alone for that particular benchmark.
“For most coding and knowledge work, start with Claude Opus 5.5 and move up to Fable only when the task needs it.”
— Anthropic, as reported in the supplied source
Limits of the Benchmark Estimates
The reported per-task costs and scores apply to Artificial Analysis’s test setup. The source does not provide enough detail here to establish how closely those averages match a company’s own prompts, usage patterns, or final invoice. Actual costs can depend on token use, task complexity and whether optional fast mode is used.
The results are dated November 10, 2026, and may change as models, APIs or benchmark methods are updated. The source gives rounded scores, so equal displayed scores at High and Xhigh do not establish that their underlying results are identical. Anthropic’s claims about model capabilities, speed and safety behavior are also not independently verified by the benchmark figures presented.
Test Settings on Real Workloads
Teams considering Fable 5.2 can compare effort settings on representative tasks, recording quality, runtime and total API cost rather than relying only on the overall index. The benchmark results point to High as a potentially useful balance in this test, but they do not establish a universal best setting.
The supplied source gives no date for a further benchmark update or any next product milestone. The immediate next step for readers is to check current API pricing and model availability, then validate the reported trade-offs against their own workloads.
Key Questions
How much does one Claude Fable 5.2 task cost?
In Artificial Analysis’s Intelligence Index test, estimated average cost ranged from $3.10 at Low effort to $26.54 at Max. These are benchmark task averages, not guaranteed prices for other workloads.
Which effort level offers the best score for the cost?
The supplied results do not establish a universal best choice. High scored 62 for $9.08 per task, while Max scored 65 for $26.54. Teams should test settings against their own quality, latency and budget needs.
Do higher effort settings have higher token prices?
The source reports the same list rates at every effort level: $8 per million input tokens and $40 per million output tokens. Estimated task costs rise because the model uses more tokens at higher effort levels.
Is the $26.54 Max figure a fixed charge?
No. It is the average API cost per task in the Artificial Analysis benchmark described by the source. A real task’s bill can differ with its token use, complexity and configuration.
What is the reported difference between Fable and Mythos?
The supplied source describes Fable as available to users with safety restrictions for certain high-risk cyber and biology requests. It says Mythos has fewer such restrictions and is limited to organizations accepted into Anthropic’s verification programs; these access details are attributed to the source’s account of Anthropic’s release.
Source: ThorstenMeyerAI.com
