Claude Opus 5 vs Opus 5.5 Effort Levels Benchmark

The resultMove from Opus 5 to Opus 5.5 at any effort level from low to xhigh and your Claude Code bill drops by roughly 40-50%. Instruction following doesn’t get worse. On my suite it got a little better.Then turn the dial to max and everything flips. Opus 5.5 at max wrote 70,320 output tokens. Opus 5 wrote 10,267. Opus 5.5 took 10 minutes 54 seconds where Opus 5 took 3 minutes 6 seconds, and it cost $2.51 against $1.13.Same ten prompts. Same effort setting. Two very different effort dials.TL;DROpus 5.5 is cheaper at low, medium, high and xhigh. Its cost ratios run 0.52-0.61x, all far outside the 1.04x run-to-run noise floor.Most of that saving is the price list, not the model. Opus 5.5’s cache reads cost 60% less than Opus 5’s, and cache traffic is 74-87% of the bill below max. Price Opus 5.5’s tokens at Opus 5’s rates and at high it comes to 0.98x Opus 5. Same cost footprint.At max, Opus 5.5 is a different product. Its output jumps to 16x its own low setting. Opus 5 only doubles…

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