Thinking levels · xAI
Does Grok 4.3 write better when it thinks harder?
Yes, and here the extra thinking barely costs anything. Setting Grok 4.3 to high lifts it 167 Elo over the default, for 1.4 times the cost and a wait that's still short. The catch is where it starts: even at its best, Grok 4.3 sits at #142 on the board.
Both settings we ran
| Setting | Elo | Step | Overall | Cost / script | Time / script | Reasoning tokens | Board rank |
|---|---|---|---|---|---|---|---|
| high | 1062 ±49 | - | 74.1 | $0.030 | 1.0 min | 3,467 | #142 |
| default (no effort flag) | 895 ±42 | - | 71.7 | $0.022 | 0.5 min | 915 | #149 |
Which setting should you use?
If you're on Grok 4.3, turn high on. There's no overlap between the two Elo ranges, which is strong evidence the gain is real, and you're paying a fraction of a cent more per script. But I wouldn't pick Grok 4.3 for scripts in the first place. Grok 4.7 (low) sits at #39 for 2.0 times the price, so I'd only keep this one if it's already wired into a pipeline and switching isn't worth it yet.
Other xAI ladders
Elo comes from comparing every pair of models on every script across the full board, so these settings sit on the same scale as the leaderboard. Cost is one full script at API list prices, and time is one accepted script, retries included. Compare this ladder with other models on the thinking-levels page, or see every xAI model side by side.