Gemma 4 31B (Reasoning) vs DeepSeek V4 Flash (high)
Results updated 11 November 2026 · In Terminus-2 and Claude Code, so part of any gap can come from the harness · How we benchmark
Verdict
DeepSeek V4 Flash (high) beats Gemma 4 31B (Reasoning) by 4 points; Gemma 4 31B (Reasoning) is cheaper and faster.
- Cost per task
- $0.015 vs $0.080
- Time per task
- 2.1 min vs 5.4 min
- Tokens per attempt
- 93k vs 634k
- Image input
- Yes vs No
Gemma 4 31B (Reasoning), Terminus-2 scores higher on GeoBenchX; DeepSeek V4 Flash (high), Claude Code scores higher on Earth-Bench V2 and GS-QA2. Gemma 4 31B (Reasoning), Terminus-2 is cheaper and faster: $0.015 against $0.080 per task, and 2.1 min against 5.4 min.
Where they differ most
By kind of work, share of tasks passed, 0 to 100 · Higher is better
| Kind of work | Gemma 4 31B (Reasoning) | DeepSeek V4 Flash (high) |
|---|---|---|
| Remote sensing and raster analysis | 59 | 72 |
| Climate and time series | 47 | 56 |
| GIS analysis | 56 | 63 |
| Change detection | 52 | 56 |
Where the tasks are set: each kind of work links to its tasks and scoresCharts for each: Gemma 4 31B (Reasoning), Terminus-2 · DeepSeek V4 Flash (high), Claude Code.
Side by side
| Metric | Gemma 4 31B (Reasoning), Terminus-2 | DeepSeek V4 Flash (high), Claude Code |
|---|---|---|
| Geospatial Agent Index: Text only | 55 | 59 |
| Geospatial Agent Index: Text + vision | No data | No image input |
| Earth-Bench V2 score | 44 | 56 |
| GeoBenchX score | 61 | 53 |
| GS-QA2 score | 61 | 69 |
| Vision agents score | No data | No image input |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 83 | 90 |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 63 | 89 |
| OGD4All score (standalone; in no index) | 85 | 77 |
| GeoNatureAgent score (standalone; in no index) | 100 | 71 |
| MultiGlobeQA score (standalone; in no index) | 30 | 67 |
| GeoAgentEval score (standalone; in no index) | 88 | 80 |
| ImpactMesh score (standalone; in no index) | No data | No data |
| Remote sensing and raster analysis score | 59 | 72 |
| Change detection score | 52 | 56 |
| GIS analysis score | 56 | 63 |
| Climate and time series score | 47 | 56 |
| Making maps score | 70 | 74 |
| Cost per task | $0.015 | $0.080 |
| Time per task (at 100 output tokens/s) | 2.1 min | 5.4 min |
| Tokens per attempt | 93k | 634k |
| Turns per attempt | 9.0 | 16.1 |
| Input price per 1M tokens | $0.17 | $0.15 |
| Output price per 1M tokens | $0.40 | $0.60 |
| Context window | 262,144 | 1,048,576 |
| Image input | Yes | No |
| Developer | DeepSeek |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, Gemma 4 31B (Reasoning), Terminus-2 or DeepSeek V4 Flash (high), Claude Code?
DeepSeek V4 Flash (high), Claude Code, 59 to 55.
Which is cheaper per task, Gemma 4 31B (Reasoning), Terminus-2 or DeepSeek V4 Flash (high), Claude Code?
Gemma 4 31B (Reasoning), Terminus-2, $0.015 to $0.080.
Which writes fewer output tokens per task, Gemma 4 31B (Reasoning), Terminus-2 or DeepSeek V4 Flash (high), Claude Code?
Gemma 4 31B (Reasoning), Terminus-2, 8k to 19k.
Which has the larger context window, Gemma 4 31B (Reasoning), Terminus-2 or DeepSeek V4 Flash (high), Claude Code?
DeepSeek V4 Flash (high), Claude Code, 1,048,576 tokens to 262,144 tokens.