Towards AITowards AIToneBench

Lab Comparison

NVIDIA Models Compared

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

All 3 ranked NVIDIA 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 853Nemotron 3 Ultra 550B (high)#63 of 87 models · $0.018 per script
Lowest Cost
$0.0028 per scriptNemotron 3 Super 120B (high)Configuration #137 · Elo 551

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

The NVIDIA model I'd use is Nemotron 3 Ultra 550B (high), at $0.018 a script and 0.7 minutes. There isn't much of a trade-off to weigh. Nemotron 3 Super 120B (high) is cheaper at $0.0028, but you'd be saving pocket change and giving up 302 Elo for it. On the Super, the high setting doesn't beat medium on Writing Elo. Their ranges overlap.

Zooming out, the Ultra is still a long way down the full board, so I'd only reach for it if I specifically needed open weights from NVIDIA. DeepSeek V4.1 Flash (max) is open too, costs about the same at $0.017, and sits at #36.

You know NVIDIA for its GPUs, and the writers at the top of this board come from other labs. The board has 3 configurations of 2 Nemotron models, all open weights, and the best of them, Nemotron 3 Ultra 550B (high), sits at #123. The smaller Super lands further down, at #133 and #137 for its two settings. So this is the cheap, open corner of the board, and I wouldn't go here first for a script.

All 3 NVIDIA 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
#123Nemotron 3 Ultra 550B (high) · open weights853 ±4166.7$0.0180.7 min61.462.270.075.567.266.866.164.674.3
#133Nemotron 3 Super 120B (medium) · open weights579 ±3962.0$0.00321.5 min53.761.666.372.261.561.965.861.658.4
#137Nemotron 3 Super 120B (high) · open weights551 ±5061.3$0.00281.2 min53.561.266.772.660.361.464.557.357.4

Effort Ladder for Nemotron 3 Super 120B

Only Nemotron 3 Super 120B ran at more than one thinking level here. Its page shows how the score, cost and time change from one level to the next.

Nemotron 3 Ultra 550B Against Other NVIDIA 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