Thinking levels · Google
Does Gemini 3.5 Flash write better when it thinks harder?
Thinking harder makes Gemini 3.5 Flash more expensive, and that's about it. High thinking lands 13 Elo below the default, a gap small enough to be noise, while each script goes from $0.142 to $0.170. For a Flash model, that's already a lot to pay per script.
Both settings we ran
| Setting | Elo | Step | Overall | Cost / script | Time / script | Reasoning tokens | Board rank |
|---|---|---|---|---|---|---|---|
| high | 1074 ±55 | - | 74.5 | $0.170 | 1.4 min | 13,480 | #141 |
| default (no effort flag) | 1087 ±42 | - | 75.2 | $0.142 | 1.4 min | 10,610 | #139 |
The 95% Elo ranges overlap: Gemini 3.5 Flash (default) runs from 1046 to 1130, and Gemini 3.5 Flash (high thinking) from 1018 to 1128.
Which setting should you use?
If you're on Gemini 3.5 Flash, keep the default: it's cheaper, the wait is about the same, and high thinking doesn't buy you a better script.
The bigger question is whether to use it at all. Even the default costs about what Gemini 3.1 Pro (default) does, $0.142 against $0.143, and that's a Pro price for a Flash score. Gemini 3.7 Flash (default) sits well above it on Elo, strong evidence rather than a coin flip, for $0.028 a script. That's the one I'd switch to.
Other Google 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 Google model side by side.