Gemini 3.1 Pro (preview) (high) vs DeepSeek V4.1 Flash (max)
Results updated 31 October 2026 · In Terminus-2 and Claude Code, so part of any gap can come from the harness · How we benchmark
Verdict
Not comparable yet: Gemini 3.1 Pro (preview) (high) is not yet ranked.
- Cost per task
- $0.22 vs $0.20
- Time per task
- 2.5 min vs 10.9 min
- Tokens per attempt
- 69k vs 1.58M
- Image input
- Yes vs Yes
Gemini 3.1 Pro (preview) (high), Terminus-2 scores higher than DeepSeek V4.1 Flash (max), Claude Code on all 2 evaluations scored for both. DeepSeek V4.1 Flash (max), Claude Code is cheaper ($0.20 against $0.22 per task), but Gemini 3.1 Pro (preview) (high), Terminus-2 is faster (2.5 min against 10.9 min).
Not yet ranked. Gemini 3.1 Pro (preview) (high), Terminus-2: no GS-QA2 score yet. Not in the Agent Index, cost and time charts until it has a score on every index evaluation. Results so far are in the side-by-side table.
Where they differ most
By kind of work, share of tasks passed, 0 to 100 · Higher is better
| Kind of work | Gemini 3.1 Pro (preview) (high) | DeepSeek V4.1 Flash (max) |
|---|---|---|
| Recognising infeasible requests | 97 | 40 |
| GIS analysis | 40 | 92 |
| Spatial statistics | 54 | 31 |
| Making maps | 56 | 72 |
Where the tasks are set: each kind of work links to its tasks and scoresCharts for each: Gemini 3.1 Pro (preview) (high), Terminus-2 · DeepSeek V4.1 Flash (max), Claude Code.
Side by side
| Metric | Gemini 3.1 Pro (preview) (high), Terminus-2 | DeepSeek V4.1 Flash (max), Claude Code |
|---|---|---|
| Geospatial Agent Index: Text only | 66 | 68 |
| Geospatial Agent Index: Text + vision | No data | No data |
| Earth-Bench V2 score | 59 | 58 |
| GeoBenchX score | 73 | 55 |
| GS-QA2 score | No data | 89 |
| Vision agents score | No data | No data |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 86 | 96 |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 95 | 100 |
| OGD4All score (standalone; in no index) | No data | 79 |
| GeoNatureAgent score (standalone; in no index) | No data | 100 |
| MultiGlobeQA score (standalone; in no index) | No data | 64 |
| GeoAgentEval score (standalone; in no index) | No data | 84 |
| Remote sensing and raster analysis score | 61 | 75 |
| GIS analysis score | 40 | 92 |
| Climate and time series score | 73 | 64 |
| Spatial statistics score | 54 | 31 |
| Making maps score | 56 | 72 |
| Recognising infeasible requests score | 97 | 40 |
| Cost per task | $0.22 | $0.20 |
| Time per task (at 100 output tokens/s) | 2.5 min | 10.9 min |
| Tokens per attempt | 69k | 1.58M |
| Turns per attempt | 6.9 | 26.0 |
| Input price per 1M tokens | $2.00 | $0.30 |
| Output price per 1M tokens | $12.00 | $1.20 |
| Context window | 1,048,576 | 1,048,576 |
| Image input | Yes | Yes |
| Developer | DeepSeek |