DeepSeek V4.1 Flash (max) vs Kimi K2.6 (Reasoning)
Results updated 31 October 2026 · In Claude Code and Kimi Code, so part of any gap can come from the harness · How we benchmark
Verdict
Kimi K2.6 (Reasoning) edges DeepSeek V4.1 Flash (max) by 1 point, and is faster; DeepSeek V4.1 Flash (max) is cheaper.
- Cost per task
- $0.20 vs $0.27
- Time per task
- 10.9 min vs 3.0 min (estimated)
- Tokens per attempt
- 1.58M vs 1.05M
- Image input
- Yes vs Yes
DeepSeek V4.1 Flash (max), Claude Code scores higher on GS-QA2; Kimi K2.6 (Reasoning), Kimi Code scores higher on Earth-Bench V2 and GeoBenchX. DeepSeek V4.1 Flash (max), Claude Code is cheaper ($0.20 against $0.27 per task), but Kimi K2.6 (Reasoning), Kimi Code is faster (3.0 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 | DeepSeek V4.1 Flash (max) | Kimi K2.6 (Reasoning) |
|---|---|---|
| Spatial statistics | 31 | 54 |
| Recognising infeasible requests | 40 | 58 |
| GIS analysis | 92 | 77 |
| Making maps | 72 | 74 |
Where the tasks are set: each kind of work links to its tasks and scoresCharts for each: DeepSeek V4.1 Flash (max), Claude Code · Kimi K2.6 (Reasoning), Kimi Code.
Side by side
| Metric | DeepSeek V4.1 Flash (max), Claude Code | Kimi K2.6 (Reasoning), Kimi Code |
|---|---|---|
| Geospatial Agent Index: Text only | 68 | 69 |
| Geospatial Agent Index: Text + vision | No data | No data |
| Earth-Bench V2 score | 58 | 63 |
| GeoBenchX score | 55 | 64 |
| GS-QA2 score | 89 | 80 |
| Vision agents score | No data | No data |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 96 | No data |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 100 | No data |
| OGD4All score (standalone; in no index) | 79 | No data |
| GeoNatureAgent score (standalone; in no index) | 100 | No data |
| MultiGlobeQA score (standalone; in no index) | 64 | No data |
| GeoAgentEval score (standalone; in no index) | 84 | No data |
| Remote sensing and raster analysis score | 75 | 75 |
| GIS analysis score | 92 | 77 |
| Climate and time series score | 64 | 63 |
| Spatial statistics score | 31 | 54 |
| Making maps score | 72 | 74 |
| Recognising infeasible requests score | 40 | 58 |
| Cost per task | $0.20 | $0.27 |
| Time per task (at 100 output tokens/s) | 10.9 min | 3.0 min (estimated) |
| Tokens per attempt | 1.58M | 1.05M |
| Turns per attempt | 26.0 | No data |
| Input price per 1M tokens | $0.30 | $0.95 |
| Output price per 1M tokens | $1.20 | $4.00 |
| Context window | 1,048,576 | 262,144 |
| Image input | Yes | Yes |
| Developer | DeepSeek | Moonshot AI |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, DeepSeek V4.1 Flash (max), Claude Code or Kimi K2.6 (Reasoning), Kimi Code?
Kimi K2.6 (Reasoning), Kimi Code, 69 to 68.
Which is cheaper per task, DeepSeek V4.1 Flash (max), Claude Code or Kimi K2.6 (Reasoning), Kimi Code?
DeepSeek V4.1 Flash (max), Claude Code, $0.20 to $0.27.
Which writes fewer output tokens per task, DeepSeek V4.1 Flash (max), Claude Code or Kimi K2.6 (Reasoning), Kimi Code?
Kimi K2.6 (Reasoning), Kimi Code, 18k to 48k.
Which has the larger context window, DeepSeek V4.1 Flash (max), Claude Code or Kimi K2.6 (Reasoning), Kimi Code?
DeepSeek V4.1 Flash (max), Claude Code, 1,048,576 tokens to 262,144 tokens.