Towards AITowards AIToneBench

Lab Comparison

MiniMax Models Compared

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

All 4 ranked MiniMax 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 1640MiniMax M3.1 Flash (max)#34 of 87 models · $0.021 per script
Lowest Cost
$0.0067 per scriptMiniMax M2 (thinking always on)Configuration #128 · Elo 758

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

Use MiniMax M3.1 Flash (max). It's the best writer here, and if you're on MiniMax M3 (thinking on), I'd switch. The 242 Elo between them is a real gap, since their ranges don't overlap, and the Flash is cheaper and faster too, at $0.021 and 4.9 minutes per script. It scores higher on every metric, by far the most on substance and visual cues. Stay on MiniMax M3 only if you need to run the model yourself: MiniMax calls the Flash open-weight, but it hasn't published the weights yet.

Against the MiniMax M2 models, the Flash's biggest gaps are on what makes a draft usable: it lands much closer to the target length, writes better visual cues and follows YouTube best practices more closely. The catch is time. MiniMax M2 (thinking always on) comes back in 0.5 minutes for $0.0067, but you give up 882 Elo for it. Bad trade. And I wouldn't choose MiniMax M2.1 (thinking always on) over MiniMax M2 for scripts. Their Elo ranges overlap; that alone doesn't establish whether either writes better.

Skip the MiniMax M2 models for scripts. MiniMax M3.1 Flash (max) leads at #72, MiniMax M3 (thinking on) follows at #99, and both older models sit far below them. Every row is cheap, so money barely enters into it here: the model you run decides the result.

All 4 MiniMax Configurations, Ranked

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. Each row is a different model. 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. * marks an estimated price. 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
#72MiniMax M3.1 Flash (max)1640 ±4377.8$0.0214.9 min72.072.483.579.675.774.181.693.280.2
#99MiniMax M3 (thinking on) · open weights1398 ±3774.8$0.042*7.6 min70.970.974.478.272.673.078.592.870.8
#127MiniMax M2.1 (thinking always on) · open weights766 ±8464.7$0.0068*0.5 min61.865.868.875.055.266.671.563.849.1
#128MiniMax M2 (thinking always on) · open weights758 ±4565.1$0.00670.5 min62.066.569.276.755.666.774.562.747.2

MiniMax M3.1 Flash Against Other MiniMax 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