Claude Sonnet 5.5 (max) vs GPT-6.1 Sol (high)
Results updated 12 November 2026 · In Claude Code and Codex, so part of any gap can come from the harness · How we benchmark
Verdict
GPT-6.1 Sol (high) edges Claude Sonnet 5.5 (max) by 1 point, and is cheaper and faster.
- Cost per task
- $0.98 vs $0.059
- Time per task
- 9.5 min (estimated) vs 0.5 min
- Tokens per attempt
- 1.47M vs 109k
- Image input
- Yes vs Yes
Claude Sonnet 5.5 (max), Claude Code scores higher on GeoBenchX and Vision agents; GPT-6.1 Sol (high), Codex scores higher on Earth-Bench V2 and GS-QA2. GPT-6.1 Sol (high), Codex is cheaper and faster: $0.059 against $0.98 per task, and 0.5 min against 9.5 min.
Where they differ most
By kind of work, share of tasks passed, 0 to 100 · Higher is better
| Kind of work | Claude Sonnet 5.5 (max) | GPT-6.1 Sol (high) |
|---|---|---|
| Climate and time series | 68 | 78 |
| Change detection | 60 | 68 |
| Making maps | 84 | 81 |
| GIS analysis | 93 | 92 |
Where the tasks are set: each kind of work links to its tasks and scoresCharts for each: Claude Sonnet 5.5 (max), Claude Code · GPT-6.1 Sol (high), Codex.
Side by side
| Metric | Claude Sonnet 5.5 (max), Claude Code | GPT-6.1 Sol (high), Codex |
|---|---|---|
| Geospatial Agent Index: Text only | 79 | 80 |
| Geospatial Agent Index: Text + vision | 82 | 82 |
| Earth-Bench V2 score | 65 | 72 |
| GeoBenchX score | 81 | 77 |
| GS-QA2 score | 92 | 93 |
| Vision agents score | 89 | 86 |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 98 | 100 |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 94 | 100 |
| OGD4All score (standalone; in no index) | 97 | No data |
| GeoNatureAgent score (standalone; in no index) | 100 | No data |
| MultiGlobeQA score (standalone; in no index) | 80 | No data |
| GeoAgentEval score (standalone; in no index) | 100 | No data |
| ImpactMesh score (standalone; in no index) | No data | No data |
| Remote sensing and raster analysis score | 84 | 84 |
| Change detection score | 60 | 68 |
| GIS analysis score | 93 | 92 |
| Climate and time series score | 68 | 78 |
| Making maps score | 84 | 81 |
| Cost per task | $0.98 | $0.059 |
| Time per task (at 100 output tokens/s) | 9.5 min (estimated) | 0.5 min |
| Tokens per attempt | 1.47M | 109k |
| Turns per attempt | 18.1 | 6.6 |
| Input price per 1M tokens | $2.00 | $2.00 |
| Output price per 1M tokens | $10.00 | $10.00 |
| Context window | 1,000,000 | 1,050,000 |
| Image input | Yes | Yes |
| Developer | Anthropic | OpenAI |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, Claude Sonnet 5.5 (max), Claude Code or GPT-6.1 Sol (high), Codex?
GPT-6.1 Sol (high), Codex, 80 to 79.
Which is cheaper per task, Claude Sonnet 5.5 (max), Claude Code or GPT-6.1 Sol (high), Codex?
GPT-6.1 Sol (high), Codex, $0.059 to $0.98.
Which writes fewer output tokens per task, Claude Sonnet 5.5 (max), Claude Code or GPT-6.1 Sol (high), Codex?
GPT-6.1 Sol (high), Codex, 2k to 52k.
Which has the larger context window, Claude Sonnet 5.5 (max), Claude Code or GPT-6.1 Sol (high), Codex?
GPT-6.1 Sol (high), Codex, 1,050,000 tokens to 1,000,000 tokens.