Every Anthropic model, side by side

Claude Opus 5.5 (max effort) sits at #1 on the whole board, so if you write scripts with AI, this is the family I'd look at first. The catch is picking one: 35 configurations of 12 models, spread over 1601 Elo. And the effort setting matters a lot here. Claude Sonnet 5.5 (max effort) lands at #3, and with the effort turned down to low, it falls to #70. Same model, one flag.

All 35 configurations

Every configuration here wrote the same 10 scripts for the same four blind judges, so the writing Elo and ranks compare across the whole board, not just inside this family. A highlighted row is a model's best configuration, and the rows under it are the same model at other effort or thinking settings. Scores run 0 to 100, and cost is one full script at the API list price of the route we ran.

Board rank	Configuration	Elo	Overall	Tone	Craft	Substance	Hook	YouTube	Flow	Slop	Length	Cues	Cost / script	Time / script
#1	Claude Opus 5.5 (max effort)	2683 ±51	90.8	90.4	90.9	89.4	90.4	90.1	89.5	93.5	97.6	89.1	$2.75	17.1 min
#2	Claude Opus 5.5 (xhigh)	2523 ±43	90.0	89.7	90.3	89.3	89.4	89.6	89.0	93.4	93.5	88.4	$0.690	4.4 min
#3	Claude Sonnet 5.5 (max effort)	2512 ±41	90.1	89.3	89.8	89.1	89.6	89.3	88.4	93.6	98.2	86.6	$1.32	12.9 min
#4	Claude Fable 5.1 (max effort)	2481 ±44	90.0	89.8	90.4	88.6	89.5	89.7	88.8	93.2	94.8	86.8	$3.15	10.4 min
#5	Claude Fable 5 (max effort + 4.8 fallback)	2396 ±43	89.5	89.8	89.6	87.9	88.7	89.8	88.7	92.9	93.2	85.3	$2.52	9.1 min
#6	Claude Fable 5 (xhigh)	2360 ±43	89.3	89.4	90.2	88.1	89.5	89.0	88.7	92.2	90.4	86.2	$1.02	3.2 min
#7	Claude Opus 5.5 (high)	2356 ±40	89.3	89.0	90.0	88.9	89.7	89.2	88.5	92.3	89.6	87.2	$0.270	1.7 min
#8	Claude Opus 5 (max effort)	2342 ±41	89.0	89.3	90.5	88.2	89.8	89.2	88.9	90.9	88.2	83.7	$0.293	2.2 min
#9	Claude Sonnet 5.5 (xhigh)	2303 ±46	88.7	88.0	89.2	88.5	88.8	88.7	87.8	92.9	89.7	85.7	$0.232	2.2 min
#10	Claude Opus 5.5 (medium)	2272 ±43	88.6	88.7	89.7	88.6	89.0	88.8	88.2	91.6	85.0	88.1	$0.169	1.1 min
#11	Claude Opus 5.5 (adaptive default)	2268 ±38	88.5	88.5	89.3	88.5	88.9	88.8	88.1	91.6	86.0	87.6	$0.172	1.0 min
#13	Claude Fable 5.1 (adaptive default)	2253 ±42	88.4	89.0	89.8	87.8	89.0	88.3	88.4	91.5	84.5	86.2	$0.648	2.1 min
#14	Claude Opus 5 (xhigh)	2253 ±38	88.4	88.6	90.2	87.1	89.1	88.0	88.4	90.5	89.1	82.7	$0.203	1.4 min
#15	Claude Opus 5.5 (low)	2245 ±37	88.4	88.0	89.0	88.0	88.4	88.0	87.7	91.0	89.9	86.9	$0.136	0.8 min
#16	Claude Opus 5 (adaptive default)	2242 ±35	88.3	88.4	90.0	87.2	89.2	87.9	88.2	90.0	88.3	83.0	$0.176	1.2 min
#17	Claude Opus 5 (high)	2229 ±35	88.2	88.4	89.8	87.0	88.7	87.5	88.0	90.2	89.4	83.4	$0.171	1.2 min
#19	Claude Fable 5 (adaptive default)	2222 ±34	88.1	88.3	89.0	86.5	88.6	88.5	87.5	90.8	89.3	83.6	$0.557	2.0 min
#20	Claude Fable 5 (medium)	2221 ±37	88.1	88.3	89.3	87.4	88.5	87.0	87.7	90.8	88.6	86.0	$0.311	0.9 min
#22	Claude Fable 5 (low)	2207 ±38	88.0	88.1	89.2	87.1	88.1	87.2	87.7	90.9	89.4	85.4	$0.305	0.9 min
#23	Claude Opus 4.8 (max effort)	2187 ±44	87.8	87.4	89.0	86.3	88.6	88.2	87.1	90.8	89.2	84.7	$0.678	5.2 min
#24	Claude Fable 5 (high)	2187 ±38	87.9	88.3	89.2	87.2	88.4	87.0	87.6	90.2	87.0	85.3	$0.319	1.0 min
#25	Claude Opus 5 (medium)	2176 ±39	87.8	87.8	89.5	86.8	88.2	87.1	87.6	89.8	89.2	82.9	$0.154	1.0 min
#29	Claude Opus 5 (low)	2160 ±39	87.6	87.8	89.6	86.6	88.2	86.8	87.7	89.2	88.2	81.8	$0.151	1.0 min
#33	Claude Sonnet 5.5 (high)	2136 ±33	87.5	86.7	88.2	87.4	87.5	87.4	86.4	91.2	89.1	85.3	$0.077	0.7 min
#46	Claude Opus 4.8 (default)	2042 ±37	86.6	86.3	88.2	85.3	87.6	85.7	86.0	88.6	88.4	83.7	$0.160	1.1 min
#48	Claude Sonnet 5 (max effort)	2034 ±48	86.4	85.8	87.2	85.2	85.2	85.5	85.2	90.1	93.0	84.1	$0.733	11.4 min
#54	Claude Sonnet 5.5 (medium)	1957 ±42	85.9	86.0	87.8	86.6	87.0	85.4	85.7	89.9	79.6	84.3	$0.059	0.5 min
#63	Claude Sonnet 5.5 (adaptive default)	1918 ±39	85.6	85.5	87.6	86.2	87.0	84.7	85.5	90.3	79.4	83.5	$0.059	0.5 min
#65	Claude Opus 4.7	1910 ±37	85.4	85.6	87.1	84.1	86.4	83.3	84.0	87.8	89.0	81.2	$0.178	1.4 min
#70	Claude Sonnet 5.5 (low)	1872 ±50	85.0	85.4	87.4	85.9	87.1	84.8	85.2	90.1	73.8	83.0	$0.056	0.4 min
#73	Claude Sonnet 5 (default)	1814 ±41	84.5	84.0	86.0	83.9	85.3	82.7	84.2	87.8	86.3	79.9	$0.091	1.3 min
#75	Claude Opus 4.6	1722 ±43	83.5	82.7	84.2	82.7	83.8	83.0	81.4	87.6	88.8	81.5	$0.395	4.7 min
#77	Claude Sonnet 4.6	1703 ±53	83.1	82.6	84.8	82.9	82.5	81.3	81.2	88.5	87.7	79.0	$0.314	5.6 min
#104	Claude Sonnet 4.5	1448 ±71	76.0	76.2	78.8	76.9	79.8	74.6	72.8	83.4	68.9	73.0	$0.145	3.2 min
#140	Claude Haiku 4.5	1081 ±48	74.8	75.6	78.4	73.3	81.2	71.5	72.5	82.6	69.7	66.1	$0.066	2.2 min

Which one should you use?

If you only care about the best script, it's Claude Opus 5.5 (max effort), but you pay for it: $2.75 and 17.1 minutes per script. Most of the time I'd use Claude Opus 5.5 (xhigh) instead. It's still #2 on the board, and the top setting costs 4.0 times as much per script for the 159 Elo between them.

If you need scripts fast, Claude Opus 5.5 (high) comes in at #7 for $0.270 and finishes in 1.7 minutes, against 17.1 for the top pick. For volume, my budget pick is Claude Sonnet 5.5 (high) at $0.077. Dropping to Claude Sonnet 5.5 (low) saves you pennies and gives up a lot of quality. One I'd skip is Claude Fable 5.1 (max effort). It costs more per script than Claude Opus 5.5 (max effort) and still ranks lower.