Towards AITowards AIToneBench

Lab Comparison

Moonshot (Kimi) Models Compared

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

All 4 ranked Moonshot (Kimi) 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 2150Kimi K3.1 Pro (max)#9 of 87 models · $0.251 per script
Lowest Cost
$0.103 per scriptKimi K2.6 (high)Configuration #103 · Elo 1341

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

For scripts you'll publish, I'd use Kimi K3.1 Pro (max). It's 34 Elo ahead of Kimi K3 (max), and their ranges overlap. Overlap alone doesn't say whether one writes better. The metrics split: it gains on substance and visual cues and gives back a little on tone, flow and YouTube best practices. It does win on cost and speed: $0.251 and 5.5 minutes per script, against $0.362 and 9.9 minutes. I'd stay on Kimi K3 (max) only if you need open weights.

If you write a lot, Kimi K3 (high) is the practical one. It drops to #45, but it costs $0.128 a script and comes back in 2.8 minutes. I'd skip Kimi K2.6 (high). At $0.103 it's only a little cheaper than Kimi K3 (high), takes longer, and sits at #103.

The best Kimi on the board is now Kimi K3.1 Pro (max), at #19 overall. Moonshot calls it open-weight, but the weights aren't published yet, so the board marks it closed for now. That's annoying if you came to Kimi to run it yourself. There are only 4 Kimi configurations here, and they're far apart: 809 Elo from top to bottom.

All 4 Moonshot (Kimi) 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
#19Kimi K3.1 Pro (max)2150 ±4282.5$0.2515.5 min77.077.888.283.479.780.885.095.484.9
#26Kimi K3 (max) · open weights2116 ±3882.2$0.3629.9 min78.778.883.583.181.282.383.494.280.0
#45Kimi K3 (high) · open weights1891 ±5080.3$0.1282.8 min77.477.982.483.279.981.182.084.576.0
#103Kimi K2.6 (high) · open weights1341 ±4674.4$0.1034.1 min69.770.375.979.473.374.076.887.965.6

Effort Ladder for Kimi K3

Only Kimi K3 ran at more than one thinking level here. Its page shows how the score, cost and time change from one level to the next.

Kimi K3.1 Pro Against Other Moonshot (Kimi) 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