Claude Fable 5.2 vs Opus 5.5 vs Sonnet 5.5 Effort Level Benchmarks

The resultEffort is Claude’s setting for how much work a model puts into each reply. Claude Code has five levels, low, medium, high, xhigh and max, and you pick one with the /effort command or the --effort flag. I ran the same ten prompts through Sonnet 5.5, Opus 5.5 and Fable 5.2 at every level.Below max, you’re mostly paying the price list. From low to xhigh a session cost $0.28-0.37 on Sonnet 5.5, $0.55-0.69 on Opus 5.5 and $1.03-1.49 on Fable 5.2. Re-priced at Opus 5.5’s rates, the three spend about the same tokens at low, medium and high, 0.88-1.06x of Opus 5.5, apart from one Sonnet 5.5 session with a partial cache miss. Fable 5.2 starts spending more at xhigh.At max, Opus 5.5 and Sonnet 5.5 jump and Fable 5.2 doesn’t. Opus 5.5 wrote 80,079 output tokens, took 12m18s and cost $2.75. Sonnet 5.5 wrote 70,405, took 8m36s and cost $1.22. Fable 5.2 wrote 30,569, took 5m43s and cost $2.46. The model with the highest prices wrote the least, finished first and cost no more than Opus…

Commenting on Bubbles has been removed. Why? If you want to talk about this post, write to the author, use the comments on their blog if it has them, or reply with a "RE:" post on your own blog. Suggestions? feedback@bubbles.town