Terminus-2 – Llama 3.3 70B (Non-reasoning) vs. Claude Code – Claude Opus 4.8 (high): geospatial agent comparison
Comparison between Terminus-2 – Llama 3.3 70B (Non-reasoning) and Claude Code – Claude Opus 4.8 (high) on the Geospatial Agent Index, each benchmark, cost, time and token use, from the same tasks, agent and limits. Methodology.
Side by side
| Metric | Terminus-2 – Llama 3.3 70B (Non-reasoning) | Claude Code – Claude Opus 4.8 (high) |
|---|---|---|
| Geospatial Agent Index | 7* | 51* (partial: attempt 1 on 105 of 320 tasks) |
| Earth-Bench V2 score | 7 | 47 |
| GeoBenchX score | 6 | 55 |
| GeoAgentBench score (guided GIS workflow; outside the index) | 8 | 93 |
| GeoAnalystBench score (guided GIS workflow; outside the index) | 0 | 100 |
| Raster and remote sensing score | 4 | 50 |
| Vector and overlay score | 0 | 100 |
| Networks and routing score | No data | No data |
| Climate and time series score | 10 | 50 |
| Spatial statistics and interpolation score | 0 | 17 |
| Mapping and cartography score | 2 | 83 |
| Recognising infeasible requests score | 13 | 38 |
| Cost per task | $0.024 | $0.447 |
| Time per task | 3.9 min | 2.2 min |
| Tokens per attempt | 65k | 197k |
| Turns per attempt | 14.9 | 8.0 |
| Attempts decided | 320 | 105 |
| Input price per 1M tokens | $0.29 | $5.00 |
| Output price per 1M tokens | $2.25 | $25.00 |
| Context window | 24,000 | No data |
| Image input | No | No data |
| Developer | Meta | Anthropic |
Charts
Partial coverage. Claude Code – Claude Opus 4.8 (high): 49 of 147 Earth-Bench V2 tasks; 56 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
First eight selected entries in metric order. Full results in table view.
Data table: Geospatial Agent Index
| Model | Creator | Geospatial Agent Index | Range while attempts are pending | Benchmarks covered | Coverage (attempts) |
|---|---|---|---|---|---|
| Terminus-2 – Llama 3.3 70B (Non-reasoning) | Meta | 7* | No pending judgements; coverage incomplete | 2 of 2 | 320 of 614 planned attempts |
Cost per task
Average cost per task (USD) · Lower is better
First eight selected entries in metric order. Full results in table view.
How to read this chart
Not shown: partial coverage. Claude Code – Claude Opus 4.8 (high).
Data table: Cost per task
| Model | Creator | Cost per task | Coverage (attempts) |
|---|---|---|---|
| Terminus-2 – Llama 3.3 70B (Non-reasoning) | Meta | $0.024 | 320 of 614 planned attempts |
Time per task
Average agent wall time per task · Lower is better
First eight selected entries in metric order. Full results in table view.
How to read this chart
Not shown: partial coverage. Claude Code – Claude Opus 4.8 (high).
Data table: Time per task
| Model | Creator | Time per task | Coverage (attempts) |
|---|---|---|---|
| Terminus-2 – Llama 3.3 70B (Non-reasoning) | Meta | 3.9 min | 320 of 614 planned attempts |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, Terminus-2 – Llama 3.3 70B (Non-reasoning) or Claude Code – Claude Opus 4.8 (high)?
Which is cheaper per task, Terminus-2 – Llama 3.3 70B (Non-reasoning) or Claude Code – Claude Opus 4.8 (high)?
Which is faster per task, Terminus-2 – Llama 3.3 70B (Non-reasoning) or Claude Code – Claude Opus 4.8 (high)?
Which has the larger context window, Terminus-2 – Llama 3.3 70B (Non-reasoning) or Claude Code – Claude Opus 4.8 (high)?
More: Terminus-2 – Llama 3.3 70B (Non-reasoning) · Claude Code – Claude Opus 4.8 (high) · all comparisons