Claude Opus 4.8 (none)

anthropic/claude-opus-4-8

Every ranked configuration of this model. Rank is dollars per unit of score on that job, and the unit is the job's own — the `per` column names it, because a rating point and a solved task are not the same thing to divide by. Displayed values are rounded; exact values in site.json and the dumps.
job	benchmark	#	score	task $	ratio $	per	released
Code	ALE-Bench	48	1411.83	$0.421746	$0.0002987	rating point	2026-05-28
Code	ALE-Bench	58	1563.83	$0.963945	$0.0006164	rating point	2026-05-28
Reason	ProofBench	18	0.69	$2.29313	$3.323	correct answer	2026-05-28
Agent	DeepSWE	18	0.408	$2.29337	$5.621	solved task	2026-05-28
Agent	DeepSWE	21	0.487	$3.44386	$7.076	solved task	2026-05-28
Agent	DeepSWE	26	0.518	$4.28194	$8.271	solved task	2026-05-28
Agent	DeepSWE	35	0.544	$8.00636	$14.73	solved task	2026-05-28
Agent	DeepSWE	41	0.59	$13.2226	$22.42	solved task	2026-05-28

These are the numbers on #frontier. How a fixed capability's cost falls over time is on /methodology.