Kimi Code – Kimi K2.6 (always on) vs. Claude Code – Claude Haiku 5.5 (high): geospatial agent comparison

Comparison between Kimi Code – Kimi K2.6 (always on) and Claude Code – Claude Haiku 5.5 (high) on the Geospatial Agent Index, each benchmark, cost, time and token use, from the same tasks, agent and limits. Methodology.

Side by side

Kimi Code – Kimi K2.6 (always on) vs. Claude Code – Claude Haiku 5.5 (high)
Metric	Kimi Code – Kimi K2.6 (always on)	Claude Code – Claude Haiku 5.5 (high)
Geospatial Agent Index	56* (partial: attempt 1 on 72 of 320 tasks)	60*
Earth-Bench V2 score	55	58
GeoBenchX score	57	62
GeoAgentBench score (guided GIS workflow; outside the index)	No data	98
GeoAnalystBench score (guided GIS workflow; outside the index)	No data	84
Raster and remote sensing score	60	56
Vector and overlay score	0	33
Networks and routing score	No data	No data
Climate and time series score	0	66
Spatial statistics and interpolation score	67	29
Mapping and cartography score	83	78
Recognising infeasible requests score	40	56
Cost per task	$0.219	$0.011
Time per task	9.7 min	1.1 min
Tokens per attempt	840k	341k
Turns per attempt	No data	8.3
Attempts decided	71	320
Input price per 1M tokens	$0.95	$0.10
Output price per 1M tokens	$4.00	$0.50
Context window	262,144	No data
Image input	Yes	No data
Developer	Moonshot AI	Anthropic

Charts

Partial coverage. Kimi Code – Kimi K2.6 (always on): 11 of 147 Earth-Bench V2 tasks; 61 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

Planned attempts still to come

Data table: Geospatial Agent Index

Geospatial Agent Index
Model	Creator	Geospatial Agent Index	Range while attempts are pending	Benchmarks covered	Coverage (attempts)
Claude Code – Claude Haiku 5.5 (high)	Anthropic	60*	No pending judgements; coverage incomplete	2 of 2	320 of 614 planned attempts

Cost per task

Average cost per task (USD) · Lower is better

Planned attempts still to come

How to read this chart

Not shown: partial coverage. Kimi Code – Kimi K2.6 (always on).

Data table: Cost per task

Cost per task
Model	Creator	Cost per task	Coverage (attempts)
Claude Code – Claude Haiku 5.5 (high)	Anthropic	$0.011	320 of 614 planned attempts

Time per task

Average agent wall time per task · Lower is better

Planned attempts still to come

How to read this chart

Not shown: partial coverage. Kimi Code – Kimi K2.6 (always on).

Data table: Time per task

Time per task
Model	Creator	Time per task	Coverage (attempts)
Claude Code – Claude Haiku 5.5 (high)	Anthropic	1.1 min	320 of 614 planned attempts

Frequently asked questions

Which scores higher on the Geospatial Agent Index, Kimi Code – Kimi K2.6 (always on) or Claude Code – Claude Haiku 5.5 (high)?

Not comparable yet: Kimi Code – Kimi K2.6 (always on) has partial coverage (11 of 147 Earth-Bench V2 tasks; 61 of 173 GeoBenchX tasks), so it is not ranked. Kimi Code – Kimi K2.6 (always on): 71 of 614 planned attempts. Claude Code – Claude Haiku 5.5 (high): 320 of 614 planned attempts. Coverage is partial; scores can change as more attempts are decided.

Which is cheaper per task, Kimi Code – Kimi K2.6 (always on) or Claude Code – Claude Haiku 5.5 (high)?

Claude Code – Claude Haiku 5.5 (high), at $0.011 against $0.219 for Kimi Code – Kimi K2.6 (always on).

Which is faster per task, Kimi Code – Kimi K2.6 (always on) or Claude Code – Claude Haiku 5.5 (high)?

Claude Code – Claude Haiku 5.5 (high), at 1.1 min against 9.7 min for Kimi Code – Kimi K2.6 (always on).

Which has the larger context window, Kimi Code – Kimi K2.6 (always on) or Claude Code – Claude Haiku 5.5 (high)?

Not known yet: Claude Code – Claude Haiku 5.5 (high) has no context window data.

More: Kimi Code – Kimi K2.6 (always on) · Claude Code – Claude Haiku 5.5 (high) · all comparisons