Gemini 3.8 Flash (low) 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
Gemini 3.8 Flash (low) beats DeepSeek V4.1 Flash (max) by 3 points, and is cheaper and faster.
- Cost per task
- $0.11 vs $0.20
- Time per task
- 1.9 min vs 10.9 min
- Tokens per attempt
- 314k vs 1.58M
- Image input
- Yes vs Yes
Gemini 3.8 Flash (low), Terminus-2 scores higher on GeoBenchX; DeepSeek V4.1 Flash (max), Claude Code scores higher on Earth-Bench V2 and GS-QA2. Gemini 3.8 Flash (low), Terminus-2 is cheaper and faster: $0.11 against $0.20 per task, and 1.9 min against 10.9 min.
Where they differ most
By kind of work, share of tasks passed, 0 to 100 · Higher is better
| Kind of work | Gemini 3.8 Flash (low) | DeepSeek V4.1 Flash (max) |
|---|---|---|
| Recognising infeasible requests | 100 | 40 |
| Spatial statistics | 62 | 31 |
| Making maps | 58 | 72 |
| Climate and time series | 54 | 64 |
Where the tasks are set: each kind of work links to its tasks and scoresCharts for each: Gemini 3.8 Flash (low), Terminus-2 · DeepSeek V4.1 Flash (max), Claude Code.
Side by side
| Metric | Gemini 3.8 Flash (low), Terminus-2 | DeepSeek V4.1 Flash (max), Claude Code |
|---|---|---|
| Geospatial Agent Index: Text only | 71 | 68 |
| Geospatial Agent Index: Text + vision | No data | No data |
| Earth-Bench V2 score | 54 | 58 |
| GeoBenchX score | 74 | 55 |
| GS-QA2 score | 86 | 89 |
| Vision agents score | No data | No data |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 98 | 96 |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 89 | 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 | 73 | 75 |
| GIS analysis score | 87 | 92 |
| Climate and time series score | 54 | 64 |
| Spatial statistics score | 62 | 31 |
| Making maps score | 58 | 72 |
| Recognising infeasible requests score | 100 | 40 |
| Cost per task | $0.11 | $0.20 |
| Time per task (at 100 output tokens/s) | 1.9 min | 10.9 min |
| Tokens per attempt | 314k | 1.58M |
| Turns per attempt | 19.0 | 26.0 |
| Input price per 1M tokens | $0.75 | $0.30 |
| Output price per 1M tokens | $3.75 | $1.20 |
| Context window | 1,048,576 | 1,048,576 |
| Image input | Yes | Yes |
| Developer | DeepSeek |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, Gemini 3.8 Flash (low), Terminus-2 or DeepSeek V4.1 Flash (max), Claude Code?
Gemini 3.8 Flash (low), Terminus-2, 71 to 68.
Which is cheaper per task, Gemini 3.8 Flash (low), Terminus-2 or DeepSeek V4.1 Flash (max), Claude Code?
Gemini 3.8 Flash (low), Terminus-2, $0.11 to $0.20.
Which writes fewer output tokens per task, Gemini 3.8 Flash (low), Terminus-2 or DeepSeek V4.1 Flash (max), Claude Code?
Gemini 3.8 Flash (low), Terminus-2, 8k to 48k.
Which has the larger context window, Gemini 3.8 Flash (low), Terminus-2 or DeepSeek V4.1 Flash (max), Claude Code?
They are level, at 1,048,576 tokens.