Gemini 3.8 Flash
Low list price, heavy usage. A cheap-looking Flash model that spends many tokens per task.
Which reasoning effort setting?
controls how long Gemini 3.8 Flash thinks before answering. Token prices never change; higher settings spend more and raise task cost.
Start at Medium. It scores 30.1 against 30.8 at the top setting, for 19% less per benchmark task. Rule: cheapest setting within 2 points of the model’s best score.
At Medium, Gemini 3.8 Flash thinks for about 47.2K tokens per task, then writes about 34.8K more as its answer: $6.26 per task, the cheapest setting.
Output rate: $3.75/1M at every effort level. Thinking tokens are billed as output.
| Effort | Index | Per task | Thinking | All output | × Medium | Time |
|---|---|---|---|---|---|---|
| MediumStart here | 30.1 | $6.26 | 47.2K | 82.1K | 1.00× | — |
| High | 30.8 | $7.77 | 60.1K | 98.9K | 1.24× | — |
Per task of the Artificial Analysis Intelligence Index: the average over its nine evaluations, counting input, cache, and output tokens at list prices. These are long, agent-style benchmark tasks, much longer than a chat message.
Where Gemini 3.8 Flash sits on the frontier
Intelligence Index against one 100K-input, 10K-output request.