🔍 Read the full analysis: Claude Fable 5.2’S #1 Ranking And The Cost Of Turning Up Effort on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Artificial Analysis ranked Anthropic’s Claude Fable 5.2 first among 74 models on its Intelligence Index at Max effort, with a score of 65. The same evaluation shows a steep cost and time trade-off: Max averaged $26.54 and 46.4 minutes per task, while High scored 62 for $9.08. The ranking is based on one benchmark, not a guarantee of better results in everyday work.
Anthropic’s Claude Fable 5.2 ranked first among 74 models on the Artificial Analysis Intelligence Index at its highest effort setting, scoring 65 at Max. The benchmark also puts a price on that result: a Max-effort task averaged $26.54 and 46.4 minutes, compared with $9.08 and 17.9 minutes at High, which scored 62.
Artificial Analysis tested five effort levels—Low, Medium, High, Xhigh and Max—on its Intelligence Index, which covers nine evaluations in areas including agentic coding, business documents, science and knowledge. Its reported cost per task is the average API cost for one task in that test. At the default Medium setting, Fable scored 56 at $3.80 per task; Low scored 54 at $3.10. High scored 62 at $9.08, and Xhigh also scored 62, at $14.83.
The results show that Max was the top-scoring setting in this index, but not that it was the most cost-effective for every use. Max cost nearly three times as much as High for three additional index points and took almost 2.6 times as long on average. On Artificial Analysis’s FrontierCode 2.0 agentic coding test, the source says Fable at High scored 49.1%, slightly above its Max result of 48.8%. These outcomes are specific to the reported tests.
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, replacing their 5.1 versions, according to the source. Anthropic says both have list prices of $8 per million input tokens and $40 per million output tokens; the effort setting affects how many tokens the model uses. The company also says answers are generated up to 30% faster than with the 5.1 models. An optional fast mode is priced at $16 per million input tokens and $80 per million output tokens, with Anthropic claiming generation speeds up to 2.5 times faster.
Claude Fable 5.2 ranks first at a higher cost
Artificial Analysis ranked Fable first among 74 models at Max effort. The score came with higher average task cost and time than High effort.
Effort changes the trade-off
Higher effort raised the Intelligence Index score, while costs and task times also increased.
| Setting | Score | Cost per task | Time per task |
|---|---|---|---|
| Medium | 56 | $3.80 | Not reported |
| High | 62 | $9.08 | 17.9 min |
| Xhigh | 62 | $14.83 | Not reported |
| Max | 65 | $26.54 | 46.4 min |
The top score has limits
Max scored highest on the Intelligence Index, but results varied by evaluation.
Max leads
Fable scored 65 at Max, three points above High for $17.46 more per task.
High edged Max
On FrontierCode 2.0, High scored 49.1%; Max scored 48.8%.
Access depends on safeguards
The source describes Fable and Mythos as the same underlying model with different restrictions and access.
- Available with safeguards for high-risk cybersecurity and most research biology tasks.
- Some blocked queries may be answered by an older model.
- Does not have those blocks.
- Limited to organizations accepted into Anthropic’s Cyber or Life Sciences Verification Programs.
Choose settings with real work
Benchmark averages may not match a company’s tasks, usage or bills.
- Compare High and Max on your own tasks.
- Track quality, token use, latency and cost.
- Treat the ranking and company claims as evidence with limits, not a guarantee of production results.
The Cost of Maximum Effort
The ranking is relevant to organizations choosing a model setting because the highest benchmark score came with substantially higher cost and latency. In the reported results, moving from Medium to High raised the score from 56 to 62 for an additional $5.28 per task. Moving from High to Xhigh added $5.75 without changing the rounded score; moving from High to Max added three points for $17.46 more per task. Actual bills will depend on usage and tasks, but the comparison makes clear that the model’s listed token prices alone do not determine the cost.
Artificial Analysis also compared Fable 5.2 at High with Anthropic’s Claude Opus 5.5 at Max: Fable scored 62 for $9.08 per task, while Opus scored 59 for $10.04. That offers a benchmark-based reason to evaluate Fable for demanding work, but it does not establish that it will outperform Opus on a particular company’s workload. Anthropic’s own guidance, as reported in the source, is to start with Opus 5.5 for most coding and knowledge work and move to Fable when a task calls for it.
Popular right now
Roborock RockMow X130H Wire-Free 4WD Robot Lawn Mower
Breville Barista Express Espresso Machine with Built-In Grinder, Black Sesame
Vitamix 5200 Blender with 64 oz. Container, Black
As an affiliate, we earn on qualifying purchases.
Two Models, Different Access
The source describes Fable 5.2 and Mythos 5.2 as the same underlying model with different safety restrictions and access. Fable is available to users with safeguards aimed at high-risk cybersecurity work and most research biology tasks. Mythos, without those blocks, is limited to organizations accepted into Anthropic’s Cyber or Life Sciences Verification Programs. The source says that when Fable blocks a query, an older model may answer instead: Opus 4.8 for cybersecurity and Opus 5 for biology.
Anthropic says early testing found about 55% fewer blocks with Fable 5.2 than Fable 5.1. The source also recounts company claims about advanced prerelease demonstrations, including a long proof effort using roughly 1,000 agents. Those examples are Anthropic’s accounts, not independent verification of routine performance. Artificial Analysis’s ranking is a separate benchmark result and should not be conflated with the company’s demonstrations.
Limits of the Benchmark Results
The ranking reflects Artificial Analysis’s Intelligence Index and its reported test setup; it does not show how Fable will perform across every production workload. The source does not supply confidence intervals, repeated-run variation or a breakdown of how much results differ between individual tasks. Scores are rounded as displayed, so the identical 62 scores for High and Xhigh do not establish that the settings perform identically in all circumstances.
It is also unclear from the material how representative the benchmark’s average task cost and time are of a particular customer’s usage. Anthropic’s claims about faster answers, fewer safety blocks and prerelease demonstrations are company statements, and the source does not say they were independently verified. No customer adoption figures or real-world business outcomes are provided.
Choosing a Setting for Real Work
Organizations considering Fable can compare High and Max on their own tasks, tracking quality, token use, latency and cost rather than relying only on the index’s top ranking. The source reports no further benchmark milestone or scheduled update. Anthropic’s stated approach is to begin with Opus 5.5 for most coding and knowledge work and use Fable when the task warrants it; access to Mythos remains subject to the company’s verification programs.
Key Questions
What ranked first on the Artificial Analysis Intelligence Index?
Claude Fable 5.2 at Max effort scored 65 and ranked first among 74 models in the results accessed November 10, 2026.
How much did a Max-effort task cost in the test?
Artificial Analysis reported an average of $26.54 per task at Max effort, compared with $9.08 at High. These are benchmark averages, not a fixed price for every real-world request.
Does Max always perform better than High?
No. Max scored higher on the overall Intelligence Index in the reported results, but High slightly exceeded Max on the source’s FrontierCode 2.0 result. Performance depends on the evaluation and task.
How are Fable 5.2 and Mythos 5.2 different?
The source describes them as the same underlying model with different safeguards and access. Fable includes restrictions on some high-risk cyber and biology requests; Mythos is available only to organizations accepted into Anthropic’s verification programs.
Source: ThorstenMeyerAI.com
