Towards AITowards AIToneBench

Lab Comparison

Meta Models Compared

Last updated , from board data generated at 17:52 UTC.

All 7 ranked Meta configurations on ToneBench, side by side: Writing Elo, Writing Score, the nine metrics, and cost and time per script. A configuration is one model at one thinking level.

Highest Writing Elo
Elo 1930Muse Wonder (max)#18 of 87 models · $0.309 per script
Lowest Cost
$0.0029 per scriptLlama 4 Scout (no thinking)Configuration #150 · Elo -303

Recommended Configurations

Ranks in these notes are configuration ranks: each setting's place among all 151 configurations. The Highest Writing Elo card above gives the model rank.

Picks and Notes From Louis

I'd use Muse Wonder (max). It's at #38 on the whole board, and nothing else from Meta comes within 200 Elo of it. It costs $0.309 and takes 2.0 minutes per script. Keep it at max: at high, the same model drops to #78, and with no effort flag to #88, mostly because those scripts run well past the target length. The catch is the price. Several models from other labs rank above it for much less, so I'd pick it when you want Meta specifically.

If you want Meta on a tight budget, I'd take Muse Spark 1.3 (medium) at $0.038 a script. Muse Wonder (max) costs 8.2 times as much. You give up a lot for that saving, though: Muse Spark sits at #95, 479 Elo behind, and turning its thinking up to high doesn't close that gap.

Llama 4 Maverick (no thinking) costs next to nothing at $0.0046, but it's 403 Elo behind even Muse Spark, so I wouldn't use it for a script I'm publishing. As for Llama 4 Scout (no thinking), it's the cheapest model here, and at #150 I wouldn't use it for scripts at all.

With Muse Wonder, Meta finally has a model that holds its own on scripts, and it's a closed one. If you're hoping an open Meta model can write your scripts, the short answer is still no. The Llama rows are the open ones, and they're much lower: Llama 4 Maverick (no thinking) is at #114, and Llama 4 Scout (no thinking) is at #150, near the very bottom of the board and 2233 Elo below Meta's best.

All 7 Meta Configurations, Grouped by Model

Every row wrote the same 10 scripts for the same four blind judges, so these Writing Elo ratings sit on the leaderboard's scale. The rank column is the configuration rank, each setting's place among all 151 configurations on the board. Rows are grouped by model: a highlighted row is a model's best configuration, and the rows under it are the same model at other thinking levels. The ± after each Elo is half the width of its 95% range, and scores run from 0 to 100. Cost is one full script at the API list price of the route used. Each name opens its scorecard, and the head-to-head comparisons cover models from different labs.

Score colors:82 and up65 to 81.9under 65

With the metric scores shown, the table scrolls sideways.

Configuration RankConfigurationWriting EloWriting ScoreCost / ScriptTime / ScriptTone & Voice MatchWriting Craft & ClaritySubstance, Accuracy & ValueHook StrengthYouTube Best PracticesContinuity & EmotionSlop ScoreLength AdherenceVisual Cue Quality
#38Muse Wonder (max)1930 ±4580.6$0.3092.0 min76.776.686.881.879.381.584.082.578.4
#78Muse Wonder (high)1584 ±5776.7$0.1490.9 min77.075.985.779.880.181.482.141.174.1
#88Muse Wonder (adaptive)1529 ±5476.2$0.1390.8 min77.375.685.280.778.980.481.437.474.8
#94Muse Spark 1.3 (high)1455 ±5275.5$0.0481.5 min73.672.480.876.678.376.380.871.064.5
#95Muse Spark 1.3 (medium)1451 ±4875.4$0.0381.1 min73.472.780.577.877.976.580.568.565.4
#114Llama 4 Maverick (no thinking) · open weights1048 ±3169.9$0.00461.7 min64.571.072.773.866.971.577.268.769.1
#150Llama 4 Scout (no thinking) · open weights-303 ±2837.6$0.00290.9 min36.344.939.141.335.136.664.218.520.9

Effort Ladders for Meta Models

Each of these models ran at more than one thinking level. Each page shows how the score, cost and time change from one level to the next.

Muse Wonder Against Other Meta Models

Each comparison lists every thinking level of both models, with Writing Elo, cost and time side by side. The line under each pair gives the Writing Elo gap between the two best configurations. A pair outside the written match-ups opens in the comparison tool.

← All comparisons