In a post on X, @_xjdr argued that Anthropic now has "absolutely no reason" to push users toward Sonnet or Opus because newer rivals, including the GPT-5.6 family and Grok 4.5, are "much better and much cheaper." The same post argued that Fable may be strong in some frontier cases but is too expensive and restrictive for most professional use. So far, the story is the claim itself and the argument it kicked off on X, not an independently verified benchmark report.