GPT-5.5 (none)

openai/gpt-5.5-20260423

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	26	1127.58	$0.073484	$0.00006517	rating point	2026-04-23
Code	ALE-Bench	38	1589.38	$0.328965	$0.000207	rating point	2026-04-23
Code	ALE-Bench	63	1942.97	$1.51713	$0.0007808	rating point	2026-04-23
Reason	ProofBench	20	0.5	$1.68179	$3.364	correct answer	2026-04-23
Agent	DeepSWE	11	0.27	$1.2002	$4.447	solved task	2026-04-23
Agent	DeepSWE	17	0.54	$2.74934	$5.093	solved task	2026-04-23
Agent	DeepSWE	25	0.644	$5.10044	$7.922	solved task	2026-04-23
Agent	DeepSWE	30	0.67	$7.22624	$10.78	solved task	2026-04-23

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