Thinking Levels · Moonshot (Kimi)
Last updated , from board data generated at 17:52 UTC.
This page compares Kimi K3 at 2 reasoning-effort settings, each tested on the same 10 scripts: the Writing Score (the 0 to 100 consensus of the judges and code), the Writing Elo (a rating from pairwise comparisons of those scores), and the cost and time of each setting. As a model, Kimi K3 ranks #12 of 87 models through its best setting. The other ranks on this page are configuration ranks: one setting's place among all 151 configurations.
On Kimi K3, more thinking helps, and you get it by leaving the effort flag off. With no effort flag, the run the board records as max spent far more reasoning tokens than the high setting and scored 225 Elo higher, with ranges that don't overlap. That's #26 on the whole board against #45. The thinking isn't free, though: 2.8 times the cost per script, and 9.9 minutes of waiting instead of 2.8.
For anything I'm going to publish, I'd run it with no flag. That puts it among the strongest open models on the board, and length is where you see the gap most: without the flag, its scripts land much closer to the target length.
The high setting makes sense when you need lots of drafts fast. It sits at #45 for $0.128 a script and comes back in 2.8 minutes. Just don't ask for high expecting a better script, because on this model that flag cuts the thinking down.
Each point is one effort setting of Kimi K3, joined from the lowest setting to the highest. Higher and further left is better: a higher score for less money or less time.
On a small screen, Table lists the same numbers as rows.
| Setting | Writing Elo | Step | Writing Score | Cost / Script | Time / Script | Reasoning Tokens | Configuration Rank |
|---|---|---|---|---|---|---|---|
| high | 1891 ±50 | -225 | 80.3 | $0.128 | 2.8 min | 3,080 | #45 |
| max (provider default) | 2116 ±38 | - | 82.2 | $0.362 | 9.9 min | 18,568 | #26 |
Writing Elo comes from comparing every pair of models on every script, which puts these settings on the same scale as the leaderboard. Step is the Writing Elo change from the provider default. Cost is one full script at API list prices, and time is the average for one accepted script, retries included. The thinking-levels page covers the other models, and the Moonshot (Kimi) family page lists every Moonshot (Kimi) model side by side.