Claude Code – Claude Haiku 4.5 (thinking 32k) vs. Claude Code – Claude Sonnet 5 (high): geospatial agent comparison
Comparison between Claude Code – Claude Haiku 4.5 (thinking 32k) and Claude Code – Claude Sonnet 5 (high) on the Geospatial Agent Index, each benchmark, cost, time and token use, from the same tasks, agent and limits. Methodology.
Side by side
| Metric | Claude Code – Claude Haiku 4.5 (thinking 32k) | Claude Code – Claude Sonnet 5 (high) |
|---|---|---|
| Geospatial Agent Index | 36* (partial: attempt 1 on 94 of 320 tasks) | 56* (partial: attempt 1 on 104 of 320 tasks) |
| Earth-Bench V2 score | 28 | 56 |
| GeoBenchX score | 45 | 56 |
| GeoAgentBench score (guided GIS workflow; outside the index) | 69 | 100 |
| GeoAnalystBench score (guided GIS workflow; outside the index) | 100 | 83 |
| Raster and remote sensing score | 33 | 52 |
| Vector and overlay score | No data | 0 |
| Networks and routing score | No data | No data |
| Climate and time series score | 33 | 67 |
| Spatial statistics and interpolation score | 40 | 50 |
| Mapping and cartography score | 60 | 80 |
| Recognising infeasible requests score | 26 | 41 |
| Cost per task | $0.155 | $0.331 |
| Time per task | 2.5 min | 2.8 min |
| Tokens per attempt | 490k | 555k |
| Turns per attempt | 14.2 | 13.1 |
| Attempts decided | 94 | 104 |
| Input price per 1M tokens | $1.00 | $2.00 |
| Output price per 1M tokens | $5.00 | $10.00 |
| Context window | 200,000 | No data |
| Image input | Yes | No data |
| Developer | Anthropic | Anthropic |
Charts
Partial coverage. Claude Code – Claude Haiku 4.5 (thinking 32k): 47 of 147 Earth-Bench V2 tasks; 47 of 173 GeoBenchX tasks; Claude Code – Claude Sonnet 5 (high): 41 of 147 Earth-Bench V2 tasks; 63 of 173 GeoBenchX tasks. Grey placeholders identify these entries in the index, cost and time charts; their height does not represent a value. Comparable bars appear when first-attempt coverage is complete. Their available results remain in the side-by-side table.
Geospatial Agent Index
Equal-weight average of the evaluation scores, 0 to 100 · Higher is better
First eight selected entries in metric order. Full results in table view.
Data table: Geospatial Agent Index
| Model | Creator | Geospatial Agent Index | Range while attempts are pending | Benchmarks covered | Coverage (attempts) |
|---|
Cost per task
Average cost per task (USD) · Lower is better
First eight selected entries in metric order. Full results in table view.
How to read this chart
Not shown: partial coverage. Claude Code – Claude Haiku 4.5 (thinking 32k), Claude Code – Claude Sonnet 5 (high).
Data table: Cost per task
| Model | Creator | Cost per task | Coverage (attempts) |
|---|
Time per task
Average agent wall time per task · Lower is better
First eight selected entries in metric order. Full results in table view.
How to read this chart
Not shown: partial coverage. Claude Code – Claude Haiku 4.5 (thinking 32k), Claude Code – Claude Sonnet 5 (high).
Data table: Time per task
| Model | Creator | Time per task | Coverage (attempts) |
|---|
Frequently asked questions
Which scores higher on the Geospatial Agent Index, Claude Code – Claude Haiku 4.5 (thinking 32k) or Claude Code – Claude Sonnet 5 (high)?
Which is cheaper per task, Claude Code – Claude Haiku 4.5 (thinking 32k) or Claude Code – Claude Sonnet 5 (high)?
Which is faster per task, Claude Code – Claude Haiku 4.5 (thinking 32k) or Claude Code – Claude Sonnet 5 (high)?
Which has the larger context window, Claude Code – Claude Haiku 4.5 (thinking 32k) or Claude Code – Claude Sonnet 5 (high)?
More: Claude Code – Claude Haiku 4.5 (thinking 32k) · Claude Code – Claude Sonnet 5 (high) · all comparisons