AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: The New #1 AI Model: Claude Fable 5.2’S Cost At Every Effort Level on ThorstenMeyerAI.com

Before you orderOffer from Amazon

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
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

Artificial Analysis ranks Anthropic’s Claude Fable 5.2 first among 74 models at Max effort, with a score of 65. Its benchmark task costs range from $3.10 at Low to $26.54 at Max, while High scores 62 for $9.08. The ranking and prices are reported by Artificial Analysis; Anthropic’s claims about model capabilities and safety have not been independently verified in the supplied material.

Anthropic’s Claude Fable 5.2 ranks first among 74 models on Artificial Analysis’ Intelligence Index when run at Max effort, according to figures accessed November 10, 2026. The benchmark’s reported cost per task ranges from $3.10 at Low to $26.54 at Max, giving businesses a cost-and-performance trade-off behind the headline ranking.

Artificial Analysis reports a score of 65 at Max effort, with Fable 5.2 also occupying the next two positions: Xhigh and High each score 62, ranked second and third. Medium, the default effort setting in Claude apps and Claude Code, scores 56 and ranks ninth; Low scores 54 and ranks 15th. The index combines nine evaluations covering areas including agentic coding, business documents, science and knowledge work.

The reported average cost per index task rises with effort: $3.10 at Low, $3.80 at Medium, $9.08 at High, $14.83 at Xhigh and $26.54 at Max. Average task time also increases, from 6.5 minutes at Low to 46.4 minutes at Max. Artificial Analysis’ figures say High reaches a score of 62 for $9.08, while Max costs nearly three times as much for a score three points higher.

Anthropic lists the same standard token prices for all five effort settings: $8 per million input tokens and $40 per million output tokens. The difference in benchmark bills reflects how many tokens are used, including tokens spent reasoning. Artificial Analysis reports 34.9 million output tokens across its index at Low and 275 million at Max. These are benchmark averages, not a promise of what any particular customer task will cost.

At a glance
reportWhen: Fable 5.2 was released November 9, 2026…
The developmentArtificial Analysis’ November 10 ranking places Claude Fable 5.2 at No. 1 at Max effort and reports sharply different benchmark costs across its five effort settings.
AI Models · Briefing

Claude Fable 5.2 Cost at Every Effort Level

Artificial Analysis ranks Fable 5.2 first of 74 models at Max effort. Its benchmark shows a steep cost increase for a modestly higher score than High.

$26.54
Max cost per index task
65
Max effort index score
74
Models ranked

Cost rises with effort

Artificial Analysis reports these average benchmark costs per index task; actual bills depend on usage and task.

Low $3.10
Medium $3.80
High $9.08
Xhigh $14.83
Max $26.54

Score versus spend

High and Xhigh share a rounded score, while Max costs substantially more for three additional points.

EffortScoreRankCost per task
Low5415th$3.10
Medium569th$3.80
High623rd$9.08
Xhigh622nd$14.83
Max651st$26.54

Choose effort for the job

The overall ranking is not a guarantee of better results on every task.

Consider High
  • Score of 62 for $9.08 per index task.
  • A lower-cost starting point for testing company workloads.
Test before Max
  • Max scores 65 for $26.54 per index task.
  • On FrontierCode 2.0, High scored 49.1% and Max 48.8%.

What the figures do—and do not—show

Benchmark

A specific index

The ranking combines nine evaluations; it does not establish the best setting for every task.

Pricing

Usage drives the bill

Anthropic lists the same standard token prices across effort levels; benchmark costs reflect tokens used, including reasoning.

Claims

Not independently verified here

The supplied material does not independently verify Anthropic’s capability, speed or safety claims.

Practical next stepCompare effort settings on your own tasks, tracking output quality alongside full API cost.
The takeaway
Max leads this benchmark, but High offers a much lower reported task cost; test both against the work you actually need done.

What the Effort Setting Costs

The ranking matters to companies weighing model capability against API spending and response time. The figures suggest that selecting the highest setting for every job may not be economical: High scores 62 at $9.08 per index task, compared with Max’s 65 at $26.54. Xhigh costs $14.83 while posting the same rounded score as High, though scores and results on a specific task may differ.

More effort does not guarantee a better result on every evaluation. In Artificial Analysis’ FrontierCode 2.0 coding test, the source reports Fable 5.2 scored 49.1% at High and 48.8% at Max. That result is specific to this test and does not establish that High is generally superior. It does underline why businesses may need to test effort settings on their own workloads rather than infer practical value from a single overall ranking.

The source also compares Fable 5.2 at High with Anthropic’s Opus 5.5 at Max: Artificial Analysis gives the former a score of 62 for $9.08 per task and the latter a score of 59 for $10.04. This is a benchmark comparison, not independent evidence that Fable will be cheaper or more effective across all uses.

Products Worth Considering

LG 27GS60QC-B Ultragear 27-inch Curved Gaming Monitor (QHD, 180Hz, 1ms, 1000R)

LG 27GS60QC-B Ultragear 27-inch Curved Gaming Monitor (QHD, 180Hz, 1ms, 1000R)

As an affiliate, we earn on qualifying purchases.

Release, Access and Safety Split

Anthropic released Fable 5.2 and Mythos 5.2 on November 9, replacing the respective 5.1 versions, according to the supplied report. It describes Mythos as the same underlying model with some safety restrictions removed, available only to organizations accepted into Anthropic’s Cyber or Life Sciences Verification Programs. Fable is available to general users and retains safeguards intended to block certain high-risk cybersecurity and biology requests.

Anthropic says the models cost $8 per million input tokens and $40 per million output tokens, and that they write answers up to 30% faster than their 5.1 predecessors. The company also offers an optional fast mode, which it says can generate tokens up to 2.5 times faster at $16 per million input tokens and $80 per million output tokens. These are company-reported specifications; the supplied material does not include independent confirmation of the speed claims.

Anthropic has also described demanding prerelease demonstrations, including a multi-agent attempt to formalize Fermat’s Last Theorem. The company says a Fable 5.2 run with about 1,000 agents proved all weighted statements after 30 hours, while an Opus 5.5 run using the same setup reached 27% by then. The source identifies these as Anthropic’s own results, based on one run per model, not independent verification or a measure of routine customer performance.

“The company says Fable 5.2 writes answers up to 30% faster than the 5.1 versions.”

— Anthropic

Limits of the Ranking and Claims

The ranking reflects Artificial Analysis’ specific nine-evaluation index; it does not establish that Fable 5.2 is best for every task, nor that its Max setting offers the best value for a particular business. The supplied material does not describe confidence intervals or establish how representative the average task costs and timings are of real customer usage.

Anthropic’s claims about safety performance, faster answers and high-profile research demonstrations remain company claims in the material provided. Anthropic says Fable 5.2 produced about 55% fewer blocks than Fable 5.1 in early testing, but the account does not give enough detail here to assess the test design or the types of requests counted. It is also unclear how Mythos access will be reviewed in practice, how many organizations qualify, or what safeguards apply to individual deployments.

The source describes Fable and Mythos as the same underlying model with different restrictions, but it does not supply technical documentation that would independently establish the extent of that relationship. The article’s original author also disclosed that the analysis was drafted with help from Claude, an Anthropic model, while stating that neither Anthropic nor Artificial Analysis reviewed or sponsored it.

Testing Costs on Real Workloads

For users considering Fable 5.2, the immediate next step is to compare effort settings on their own tasks, tracking both output quality and full API cost. The benchmark suggests High may be a less expensive starting point than Max for some work, but it does not predict the result for a company’s specific coding, research or document tasks. Anthropic’s stated advice is to begin with Opus 5.5 for most coding and knowledge work and use Fable when greater capability is needed.

Further clarity will depend on additional independent evaluations, more detailed reporting on the benchmark’s methods and real-world usage, and Anthropic’s disclosures about safety testing and access to Mythos. The supplied material does not identify a scheduled follow-up benchmark or a date for further data.

Key Questions

What is Claude Fable 5.2’s top ranking?

Artificial Analysis ranks it first of 74 on its Intelligence Index at Max effort, with a score of 65 in figures accessed November 10, 2026.

How much does each effort level cost in the benchmark?

Artificial Analysis reports an average cost per index task of $3.10 at Low, $3.80 at Medium, $9.08 at High, $14.83 at Xhigh and $26.54 at Max. Actual costs depend on usage and the task.

Does Max effort always perform better?

No. Max has the highest overall index score in this comparison, but the supplied source says Fable 5.2 scored 49.1% at High and 48.8% at Max on Artificial Analysis’ FrontierCode 2.0 test.

Who can use Claude Mythos 5.2?

According to the supplied report, Mythos is limited to organizations accepted into Anthropic’s Cyber or Life Sciences Verification Programs. The source does not specify how many organizations have been accepted.

Are the model’s reported research achievements independently verified?

Not in the supplied material. The demonstrations, including the multi-agent Fermat’s Last Theorem run, are described as Anthropic’s own results, not independently verified findings.

Source: ThorstenMeyerAI.com

You May Also Like

Will Robots Share The Housework, Or Change Who Does It?

The Information reported that Meta and Chinese firm Fourier built a humanoid prototype. Meta says the work is internal research, not a product launch.

The Coding And Cowork Questions Behind Qwen3.8-Max

An IdeaNavigator AI brief frames Qwen3.8-Max as a test case for role-filtered AI monitoring, not as a verified model launch or benchmark report.

Claude Haiku 5.5 Joins The AWS AI Lineup

An AWS page titled “Introducing Claude Haiku 5.5 on AWS” names the model, but the available material does not confirm launch timing, access or pricing.

Could You Join Anthropic As A Guest Investigator In The Life Sciences?

Anthropic has invited guest investigators in the life sciences, but available details do not explain eligibility, responsibilities, application steps or timing.