Claude Haiku 4.5 (Reasoning) vs DeepSeek V4 Flash (high)
Results updated 30 October 2026 · Both in Claude Code · How we benchmark
Verdict
Claude Haiku 4.5 (Reasoning) edges DeepSeek V4 Flash (high) by 2 points, and is faster; DeepSeek V4 Flash (high) is cheaper.
- Cost per task
- $0.092 vs $0.080
- Time per task
- 1.6 min vs 5.4 min
- Tokens per attempt
- 313k vs 634k
- Image input
- Yes vs No
Claude Haiku 4.5 (Reasoning), Claude Code scores higher on Earth-Bench V2 and GeoBenchX; they are level on GS-QA2. DeepSeek V4 Flash (high), Claude Code is cheaper ($0.080 against $0.092 per task), but Claude Haiku 4.5 (Reasoning), Claude Code is faster (1.6 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 | Claude Haiku 4.5 (Reasoning) | DeepSeek V4 Flash (high) |
|---|---|---|
| Recognising infeasible requests | 53 | 35 |
| Spatial statistics | 23 | 38 |
| Making maps | 56 | 68 |
| GIS analysis | 67 | 62 |
Where the tasks are set: each kind of work links to its tasks and scoresCharts for each: Claude Haiku 4.5 (Reasoning), Claude Code · DeepSeek V4 Flash (high), Claude Code.
Side by side
| Metric | Claude Haiku 4.5 (Reasoning), Claude Code | DeepSeek V4 Flash (high), Claude Code |
|---|---|---|
| Geospatial Agent Index: Text only | 58 | 56 |
| Geospatial Agent Index: Text + vision | 48 | No image input |
| Earth-Bench V2 score | 51 | 50 |
| GeoBenchX score | 53 | 50 |
| GS-QA2 score | 69 | 69 |
| Vision agents score | 19 | No image input |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 78 | 84 |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 74 | 89 |
| OGD4All score (standalone; in no index) | 95 | 70 |
| GeoNatureAgent score (standalone; in no index) | 93 | 71 |
| MultiGlobeQA score (standalone; in no index) | 70 | 64 |
| GeoAgentEval score (standalone; in no index) | 88 | 80 |
| Remote sensing and raster analysis score | 65 | 67 |
| GIS analysis score | 67 | 62 |
| Climate and time series score | 49 | 53 |
| Spatial statistics score | 23 | 38 |
| Making maps score | 56 | 68 |
| Recognising infeasible requests score | 53 | 35 |
| Cost per task | $0.092 | $0.080 |
| Time per task (at 100 output tokens/s) | 1.6 min | 5.4 min |
| Tokens per attempt | 313k | 634k |
| Turns per attempt | 10.6 | 16.1 |
| Input price per 1M tokens | $1.00 | $0.44 |
| Output price per 1M tokens | $5.00 | $1.32 |
| Context window | 200,000 | 1,048,576 |
| Image input | Yes | No |
| Developer | Anthropic | DeepSeek |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, Claude Haiku 4.5 (Reasoning), Claude Code or DeepSeek V4 Flash (high), Claude Code?
Claude Haiku 4.5 (Reasoning), Claude Code, 58 to 56.
Which is cheaper per task, Claude Haiku 4.5 (Reasoning), Claude Code or DeepSeek V4 Flash (high), Claude Code?
DeepSeek V4 Flash (high), Claude Code, $0.080 to $0.092.
Which writes fewer output tokens per task, Claude Haiku 4.5 (Reasoning), Claude Code or DeepSeek V4 Flash (high), Claude Code?
Claude Haiku 4.5 (Reasoning), Claude Code, 7k to 19k.
Which has the larger context window, Claude Haiku 4.5 (Reasoning), Claude Code or DeepSeek V4 Flash (high), Claude Code?
DeepSeek V4 Flash (high), Claude Code, 1,048,576 tokens to 200,000 tokens.