Claude Code – Claude Haiku 4.5 (thinking 32k) vs. Codex – GPT-6.1 Sol (high): geospatial agent comparison

Comparison between Claude Code – Claude Haiku 4.5 (thinking 32k) and Codex – GPT-6.1 Sol (high) on the Geospatial Agent Index, each benchmark, cost, time and token use, from the same tasks, agent and limits. Methodology.

Side by side

Claude Code – Claude Haiku 4.5 (thinking 32k) vs. Codex – GPT-6.1 Sol (high)
Metric	Claude Code – Claude Haiku 4.5 (thinking 32k)	Codex – GPT-6.1 Sol (high)
Geospatial Agent Index	36* (partial: attempt 1 on 94 of 320 tasks)	63*
Earth-Bench V2 score	28	58
GeoBenchX score	45	67
GeoAgentBench score (guided GIS workflow; outside the index)	69	100
GeoAnalystBench score (guided GIS workflow; outside the index)	100	100
Raster and remote sensing score	33	54
Vector and overlay score	No data	50
Networks and routing score	No data	No data
Climate and time series score	33	66
Spatial statistics and interpolation score	40	29
Mapping and cartography score	60	78
Recognising infeasible requests score	26	67
Cost per task	$0.155	$0.081
Time per task	2.5 min	1.8 min
Tokens per attempt	490k	149k
Turns per attempt	14.2	7.8
Attempts decided	94	320
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	OpenAI

Charts

Partial coverage. Claude Code – Claude Haiku 4.5 (thinking 32k): 47 of 147 Earth-Bench V2 tasks; 47 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)
Codex – GPT-6.1 Sol (high)	OpenAI	63*	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. Claude Code – Claude Haiku 4.5 (thinking 32k).

Data table: Cost per task

Cost per task
Model	Creator	Cost per task	Coverage (attempts)
Codex – GPT-6.1 Sol (high)	OpenAI	$0.081	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. Claude Code – Claude Haiku 4.5 (thinking 32k).

Data table: Time per task

Time per task
Model	Creator	Time per task	Coverage (attempts)
Codex – GPT-6.1 Sol (high)	OpenAI	1.8 min	320 of 614 planned attempts

Frequently asked questions

Which scores higher on the Geospatial Agent Index, Claude Code – Claude Haiku 4.5 (thinking 32k) or Codex – GPT-6.1 Sol (high)?

Not comparable yet: Claude Code – Claude Haiku 4.5 (thinking 32k) has partial coverage (47 of 147 Earth-Bench V2 tasks; 47 of 173 GeoBenchX tasks), so it is not ranked. Claude Code – Claude Haiku 4.5 (thinking 32k): 94 of 614 planned attempts. Codex – GPT-6.1 Sol (high): 320 of 614 planned attempts. Coverage is partial; scores can change as more attempts are decided.

Which is cheaper per task, Claude Code – Claude Haiku 4.5 (thinking 32k) or Codex – GPT-6.1 Sol (high)?

Codex – GPT-6.1 Sol (high), at $0.081 against $0.155 for Claude Code – Claude Haiku 4.5 (thinking 32k).

Which is faster per task, Claude Code – Claude Haiku 4.5 (thinking 32k) or Codex – GPT-6.1 Sol (high)?

Codex – GPT-6.1 Sol (high), at 1.8 min against 2.5 min for Claude Code – Claude Haiku 4.5 (thinking 32k).

Which has the larger context window, Claude Code – Claude Haiku 4.5 (thinking 32k) or Codex – GPT-6.1 Sol (high)?

Not known yet: Codex – GPT-6.1 Sol (high) has no context window data.

More: Claude Code – Claude Haiku 4.5 (thinking 32k) · Codex – GPT-6.1 Sol (high) · all comparisons