Vibeleaderboard
← All Intel
Intel / post

Why Claude Opus 5.5's cost per task barely moved despite price cuts

Source
Artificial Analysis
Date
From the Daily Brief

When Opus 5.5 launched, the story was lower list prices. Artificial Analysis has now measured what those prices mean per task. On its Intelligence Index, a 20% token price cut and cheaper cache reads nearly offset an increase of about 80% in tokens used per task, leaving cost per task close to Opus 5. On its Coding Agent Index the picture is sharper: Opus 5.5 set a new high score of 66, with gains on Terminal-Bench, DeepSWE and SWE-Atlas-QnA, but cost per task rose 21% to $13.04, driven by roughly 2.4 times more output tokens. The practical lesson is that price sheets are a poor guide to spend on reasoning models. Teams planning a switch should measure tokens per completed task on their own workloads before assuming the upgrade saves money.

Read the 2026-09-28 Brief →
Artificial Analysis@ArtificialAnlys
Thread · 2 parts

Claude Opus 5.5 (max) costs $5.98 per Intelligence Index task, which is similar to Opus 5 (max) at $5.86, but this bundles a significant token usage increase with Anthropic’s price reductions Compared to Opus 5, Opus 5.5’s increased token usage would drive an ~80% increase in cost per task to $10.51. However, the 20% cut to base prices takes that to $8.41, and then cutting cache reads further to $0.20 takes it to $5.98 (an additional 29% cut)

For more analysis of the Opus 5.5 release see https://t.co/dIBD9GMiu0

Context

Artificial Analysis reports that Claude Opus 5.5 (max) costs $5.98 per Intelligence Index task, similar to Opus 5 (max) at $5.86, even though Anthropic cut prices. The firm says the figure bundles a significant increase in usage with the price reductions. By its arithmetic, the extra token usage alone would raise cost per task about 80% to $10.51. A 20% cut to base prices brings that to $8.41, and cutting cache reads to $0.20 brings it to $5.98. On this accounting the price cuts offset the added token usage, leaving cost near where Opus 5 started.

Terms in this piece · Glossary
  • token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
More from Artificial Analysis
Recommended reads
Comments

Checking sign-in…

Loading comments…