Gemma 4 26B A4B (Reasoning), Terminus-2 vs. GPT-5.6 Sol (medium), Codex
Gemma 4 26B A4B (Reasoning), Terminus-2 against GPT-5.6 Sol (medium), Codex: 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 | Gemma 4 26B A4B (Reasoning), Terminus-2 | GPT-5.6 Sol (medium), Codex |
|---|---|---|
| Geospatial Agent Index: Text only | 56 | 100, not every task attempted yet (19 of 488 tasks attempted) |
| Geospatial Agent Index: Text + vision | Still to run: Vision agents | Still to run: Earth-Bench V2, GeoBenchX, GS-QA2, Vision agents |
| Earth-Bench V2 score | 46 | No data |
| GeoBenchX score | 63 | No data |
| GS-QA2 score | 58 | 100 |
| Vision agents score | No data | No data |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 86 | No data |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 63 | No data |
| Raster and remote sensing score | 59 | 100 |
| Vector and overlay score | 57 | No data |
| Climate and time series score | 47 | No data |
| Spatial statistics and interpolation score | 31 | No data |
| Mapping and cartography score | 70 | No data |
| Recognising infeasible requests score | 63 | No data |
| Cost per task | $0.021 | $0.060 |
| Time per task (at 100 output tokens/s) | 5.6 min | 0.2 min |
| Tokens per attempt | 168k | 66k |
| Turns per attempt | 14.6 | 6.1 |
| Attempts scored | 488 | 19 |
| Input price per 1M tokens | $0.10 | $4.00 |
| Output price per 1M tokens | $0.30 | $20.00 |
| Developer | OpenAI |
* Not every task attempted yet: the score may still change. What the asterisk means.
Charts
Not yet ranked. GPT-5.6 Sol (medium), Codex: Earth-Bench V2 not started; GeoBenchX not started; 19 of 221 GS-QA2 tasks attempted. Grey placeholders mark these entries in the index, cost and output token charts; their height is not a value. Bars appear once every task has an attempt. Results so far are in the side-by-side table.
Geospatial Agent Index
Equal average of Earth-Bench V2, GeoBenchX and GS-QA2, 0 to 100 · Higher is better
Top eight. Every model is in the table.
Data table: Geospatial Agent Index
| Model | Developer | Geospatial Agent Index | Status | Evaluations scored | Tasks attempted |
|---|---|---|---|---|---|
| Gemma 4 26B A4B (Reasoning), Terminus-2 | 56 | Complete | 3 of 3 | 488 of 488 tasks |
Cost per task
Average cost per task (USD) · Lower is better
Top eight. Every model is in the table.
How to read this chart
Not shown: not yet ranked. 1 entries without an attempt on every index task. What is still to run is on the models page.
Data table: Cost per task
| Model | Developer | Cost per task | Tasks attempted |
|---|---|---|---|
| Gemma 4 26B A4B (Reasoning), Terminus-2 | $0.021 | 488 of 488 tasks |
Output tokens per task
Average output tokens per attempt, reasoning included · Lower is better
Top eight. Every model is in the table.
How to read this chart
Not shown: not yet ranked. 1 entries without an attempt on every index task. What is still to run is on the models page.
Data table: Output tokens per task
| Model | Developer | Output tokens per task | Tasks attempted |
|---|---|---|---|
| Gemma 4 26B A4B (Reasoning), Terminus-2 | 21k | 488 of 488 tasks |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, Gemma 4 26B A4B (Reasoning), Terminus-2 or GPT-5.6 Sol (medium), Codex?
Which is cheaper per task, Gemma 4 26B A4B (Reasoning), Terminus-2 or GPT-5.6 Sol (medium), Codex?
Which writes fewer output tokens per task, Gemma 4 26B A4B (Reasoning), Terminus-2 or GPT-5.6 Sol (medium), Codex?
More: Gemma 4 26B A4B (Reasoning), Terminus-2 · GPT-5.6 Sol (medium), Codex · All comparisons