
Fable 5.2 cost notes and tier scores
The top tier is not automatically the strongest. Self-reported tests suggest starting with medium; Max costs more and is not necessarily better.
Brief
This item merges the official announcement and two review posts. @claudeai announced Fable 5.2 and Mythos 5.2. @DDJCXX retells official coding-test scores: Medium 37.6%, High 45.3%, Xhigh 49.1%, and Max 45.6%, but Max costs more than three times as much as Xhigh, so the top writing-code tier is not actually the strongest. The scores and the price comparison come from the author's retelling of official material, should be checked against the official notes, and have not been independently verified.
@wquguru says Opus 5.5 medium is the daily driver and Fable 5.2 is the advisor, at about twice the unit price of Opus; a routine Python bug costs about $1.87 on Fable medium and about $0.97 on Opus, with the gap at the high tier around 4.5×. Those costs are the author's claims and have not been verified. The usage is /advisor fable, and you can also switch with /effort and /advisor off.
Why it is useful
There are comparable numbers for choosing a model tier. Start with medium, and do not jump to Max by default.
Sources (3 original posts)
-
View the original post on X ↗
Claude@claudeai
https://x.com/claudeai/status/2119867114450268322 -
View the original post on X ↗
硅基废话@DDJCXX
https://x.com/DDJCXX/status/2119972412204994917 -
View the original post on X ↗
wquguru@wquguru
https://x.com/wquguru/status/2120366072910282765
This is a brief, not the original postThis page is an English brief, not a full reprint. Check the original post for details.