Gemini 4 Neon
Multimodal knowledge work. Complex coding, analytical reasoning and mixed-media workflows.
Which reasoning effort setting?
controls how long Gemini 4 Neon thinks before answering. Token prices never change; higher settings spend more and raise task cost.
Start at High. It scores 59.6 against 61.4 at the top setting, for 42% less per benchmark task. Rule: cheapest setting within 2 points of the model’s best score.
At High, Neon thinks for about 46.3K tokens per task, then writes about 43.9K more as its answer: $3.74 per task, 5.0× Low, for +15.5 points.
Output rate: $15.00/1M at every effort level. Thinking tokens are billed as output.
| Effort | Index | Per task | Thinking | All output | × Low | Time |
|---|---|---|---|---|---|---|
| Low | 44.1 | $0.75 | 8.9K | 21.7K | 1.00× | 3.7 min |
| Medium | 54.9 | $1.91 | 23.9K | 50K | 2.56× | 8.3 min |
| HighStart here | 59.6 | $3.74 | 46.3K | 90.2K | 5.02× | 14.8 min |
| Xhigh | 61.4 | $6.45 | 80.5K | 142K | 8.65× | 21.5 min |
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 4 Neon sits on the frontier
Intelligence Index against one 100K-input, 10K-output request.