AI Agent StoreLow
Tested setting0.0 pts vs Low
3,331 output tokens
Google · REASONING EFFORT GUIDEGemini 3.1 Pro Preview
Gemini 3.1 Pro Preview
How much thinking?
Compare Low vs Medium where tested. Find the effort setting that fits your task, and see how much extra reasoning changes the result.
SAME MODEL. DIFFERENT THINKING.
Choose a thinking level for Gemini 3.1 Pro Preview
OckBench
Correct answers out of 200 problems. Domain proportions are 100 math, 60 coding and 40 science.
200 selected problems · Source release date not specified · Reviewed 2026-10-15Check the source ↗SUGGESTED SETTING TO TESTMedium
Lowest measured token use within 1.0 points of the best tested overall accuracy.
87.5 / 100 · 7,116 output tokensExact best1.0 pts10 pts
Set 0 for the highest score. Increase the gap to explore cheaper or lighter options.
THE EFFORT CURVE
Does more thinking help?
● Near-best pickOrange = score decreased
Scores out of 100 · Full 0–100 scale · tap a point to select mode B. Cyan band: within 1.0 points of the best tested score.
WHAT YOU SPEND
Tokens used to answer
Source-reported average output tokens, including reasoning, for the selected scope.
Low3,331
MediumNear-best pick7,116
Measured in this study with its particular tools and setup. Your task mix, tools, caching and endpoint can change the result.
Score difference · B minus A+8.0 pts
Large gap · B scored higher
Highest minus lowest score in this study8.0 pts
Best minus lowest tested score
Highest observed score87.5
Medium
We describe gaps under 1 point as small, 1–under 5 as noticeable, and 5 or more as large. These are reading aids, not proof that a difference would hold up in repeated tests. Equal scores can still hide different mistakes.
EVERY TESTED MODE
2 modes · untested settings are left blankResults for each tested setting
79.5 / 100
Medium
Suggested setting+8.0 pts vs Low
7,116 output tokens
Full benchmark table & exact configuration IDs
Swipe inside the table for all measures. Figures from different study selections are not directly comparable.
| Effort | Overall accuracy | Mathematics | Coding tests | Science | Cost / task | Median seconds | Output tokens | Exact source ID |
|---|---|---|---|---|---|---|---|---|
| Low | 79.5 | 73.0 | 91.7 | 77.5 | — | — | 3,331 | gemini-3.1-pro-preview-low |
| Medium | 87.5 | 87.0 | 90.0 | 85.0 | — | — | 7,116 | gemini-3.1-pro-preview-medium |
SCALE THE OBSERVED COST
Effort cost difference calculator
What would A versus B cost if you repeated this benchmark’s average workload?
A · Low—
B · Medium—
Difference—This study did not report comparable costs
This source has no comparable dollar-cost observation. Use the token calculator below with your own usage; token counts alone cannot supply a full bill.
YOUR WORKLOAD
Reasoning-token cost calculator
Estimate the extra bill from thinking tokens. The starting token counts are editable examples, not measurements of Gemini 3.1 Pro Preview. Rates start from the current model snapshot.
A · Low$44estimated monthly cost
B · Medium$92estimated monthly cost
Mode B costs $48 more per month with these assumptions.
This estimate uses the token counts and prices you enter. Excludes retries, tools, caching, long-context premiums, hosting and taxes. Includes thinking in output billing; do not also include it in “visible output.” Check endpoint limits separately. Changing mode does not automatically predict tokens or quality.
Gemini 3.1 Pro Preview effort benchmarks at a glance
Each card compares settings tested together in one study. Scores from different benchmarks or question sets measure different work and are not averaged together.
Math, code & science · OckBench
Correct answers out of 200 problems. Domain proportions are 100 math, 60 coding and 40 science.
- Low
- 79.5
- Medium · pick
- 87.5
Near-best pick: Medium within 1 point of the best score. Highest-to-lowest score gap: 8.0 points.
Explore this study ↗All results and the exact settings tested
A dash means the study did not report that result. This table compares only the versions and settings tested on the same questions in this study.
| Effort / budget | Overall accuracy | Mathematics | Coding tests | Science | Average output tokens | Exact source configuration |
|---|---|---|---|---|---|---|
| Low | 79.5 | 73.0 | 91.7 | 77.5 | 3,331 | gemini-3.1-pro-preview-low |
| Medium | 87.5 | 87.0 | 90.0 | 85.0 | 7,116 | gemini-3.1-pro-preview-medium |
Reasoning effort, explained
Which Gemini 3.1 Pro Preview thinking level is most efficient?
Medium is the lowest-token option within 1 point of the best overall accuracy in OckBench. The tested score range is 8.0 points. This allows a score up to 1 point below the best result in the first listed study. Change the study, task and allowed score gap above to explore another comparison. It is a shortlist for testing your own workload.
Which Gemini 3.1 Pro Preview effort settings are compared?
Low, Medium appear in the collected studies. Each chart only includes modes tested together in that study. These published labels do not establish that every endpoint or application exposes every setting.
Does more effort improve Gemini 3.1 Pro Preview?
In the first listed OckBench sweep, the tested score range is 8.0 points and no adjacent effort increase lowers the score. This does not guarantee a gain on every task or with a different testing setup.
How much more does high effort cost for Gemini 3.1 Pro Preview?
Use the A/B selector and measured-cost calculator when the source reports dollar costs. For your own agent, enter average input, visible output and thinking tokens in the token calculator. Higher effort is adaptive, so the label itself cannot tell you the bill. Missing costs stay missing.