gpt-oss-120b (high), Codex vs. Claude Opus 5.5 (low), Claude Code
gpt-oss-120b (high), Codex against Claude Opus 5.5 (low), Claude Code: Geospatial Agent Index, each evaluation, cost, time and tokens, on the same tasks and limits. The harnesses differ, so part of any gap can come from the harness. Methodology.
Side by side
| Metric | gpt-oss-120b (high), Codex | Claude Opus 5.5 (low), Claude Code |
|---|---|---|
| Geospatial Agent Index: Text only | 43 | 74 |
| Geospatial Agent Index: Text + vision | No image input | 78 |
| Earth-Bench V2 score | 49 | 59 |
| GeoBenchX score | 55 | 77 |
| GS-QA2 score | 25 | 86 |
| Vision agents score | No image input | 89 |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 58 | 96 |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 42 | 100 |
| OGD4All score (standalone; in no index) | No data | 95 |
| GeoNatureAgent score (standalone; in no index) | No data | 100 |
| MultiGlobeQA score (standalone; in no index) | No data | 79 |
| GeoAgentEval score (standalone; in no index) | No data | 100 |
| Remote sensing and raster analysis score | 40 | 79 |
| GIS analysis score | 23 | 84 |
| Climate and time series score | 44 | 59 |
| Spatial statistics score | 0 | 77 |
| Making maps score | 61 | 77 |
| Recognising infeasible requests score | 62 | 77 |
| Cost per task | $0.26 | $0.081 |
| Time per task (at 100 output tokens/s) | 6.0 min | 0.4 min |
| Tokens per attempt | 700k | 63k |
| Turns per attempt | 29.8 | 3.5 |
| Input price per 1M tokens | $0.35 | $4.00 |
| Output price per 1M tokens | $0.75 | $20.00 |
| Context window | 128,000 | 1,000,000 |
| Image input | No | Yes |
| Developer | OpenAI | Anthropic |
Charts
Geospatial Agent Index
Equal average of Earth-Bench V2, GeoBenchX and GS-QA2, 0 to 100 · Higher is better
Data table: Geospatial Agent Index
| Model | Developer | Geospatial Agent Index |
|---|---|---|
| Claude Opus 5.5 (low), Claude Code | Anthropic | 74 |
| gpt-oss-120b (high), Codex | OpenAI | 43 |
Cost per task
Average cost per task (USD) · Lower is better
Data table: Cost per task
| Model | Developer | Cost per task |
|---|---|---|
| Claude Opus 5.5 (low), Claude Code | Anthropic | $0.081 |
| gpt-oss-120b (high), Codex | OpenAI | $0.26 |
Output tokens per task
Average output tokens per attempt, reasoning included · Lower is better
Data table: Output tokens per task
| Model | Developer | Output tokens per task |
|---|---|---|
| Claude Opus 5.5 (low), Claude Code | Anthropic | 1k |
| gpt-oss-120b (high), Codex | OpenAI | 27k |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, gpt-oss-120b (high), Codex or Claude Opus 5.5 (low), Claude Code?
Claude Opus 5.5 (low), Claude Code, 74 to 43.
Which is cheaper per task, gpt-oss-120b (high), Codex or Claude Opus 5.5 (low), Claude Code?
Claude Opus 5.5 (low), Claude Code, $0.081 to $0.26.
Which writes fewer output tokens per task, gpt-oss-120b (high), Codex or Claude Opus 5.5 (low), Claude Code?
Claude Opus 5.5 (low), Claude Code, 1k to 27k.
Which has the larger context window, gpt-oss-120b (high), Codex or Claude Opus 5.5 (low), Claude Code?
Claude Opus 5.5 (low), Claude Code, 1,000,000 tokens to 128,000 tokens.
More: gpt-oss-120b (high), Codex · Claude Opus 5.5 (low), Claude Code · All comparisons