Mistral Large 4 (high) vs Nemotron 3.5 Super VL 120B (Reasoning)
Results updated 3 November 2026 · Both in Terminus-2 · How we benchmark
Verdict
Mistral Large 4 (high) beats Nemotron 3.5 Super VL 120B (Reasoning) by 7 points; Nemotron 3.5 Super VL 120B (Reasoning) is cheaper and faster.
- Cost per task
- $0.10 vs $0.0096
- Time per task
- 12.0 min vs 3.8 min
- Tokens per attempt
- 226k vs 98k
- Image input
- Yes vs Yes
Mistral Large 4 (high), Terminus-2 scores higher on GeoBenchX and GS-QA2; they are level on Earth-Bench V2. Nemotron 3.5 Super VL 120B (Reasoning), Terminus-2 is cheaper and faster: $0.0096 against $0.10 per task, and 3.8 min against 12.0 min.
Where they differ most
By kind of work, share of tasks passed, 0 to 100 · Higher is better
| Kind of work | Mistral Large 4 (high) | Nemotron 3.5 Super VL 120B (Reasoning) |
|---|---|---|
| GIS analysis | 82 | 51 |
| Recognising infeasible requests | 23 | 43 |
| Making maps | 82 | 67 |
| Spatial statistics | 46 | 31 |
Where the tasks are set: each kind of work links to its tasks and scoresCharts for each: Mistral Large 4 (high), Terminus-2 · Nemotron 3.5 Super VL 120B (Reasoning), Terminus-2.
Side by side
| Metric | Mistral Large 4 (high), Terminus-2 | Nemotron 3.5 Super VL 120B (Reasoning), Terminus-2 |
|---|---|---|
| Geospatial Agent Index: Text only | 59 | 52 |
| Geospatial Agent Index: Text + vision | No data | No data |
| Earth-Bench V2 score | 46 | 46 |
| GeoBenchX score | 55 | 53 |
| GS-QA2 score | 77 | 59 |
| Vision agents score | No data | No data |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 80 | 68 |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 94 | 37 |
| OGD4All score (standalone; in no index) | 96 | 91 |
| GeoNatureAgent score (standalone; in no index) | 100 | 92 |
| MultiGlobeQA score (standalone; in no index) | 62 | 73 |
| GeoAgentEval score (standalone; in no index) | 60 | 80 |
| Remote sensing and raster analysis score | 67 | 61 |
| GIS analysis score | 82 | 51 |
| Climate and time series score | 40 | 46 |
| Spatial statistics score | 46 | 31 |
| Making maps score | 82 | 67 |
| Recognising infeasible requests score | 23 | 43 |
| Cost per task | $0.10 | $0.0096 |
| Time per task (at 100 output tokens/s) | 12.0 min | 3.8 min |
| Tokens per attempt | 226k | 98k |
| Turns per attempt | 13.8 | 10.1 |
| Input price per 1M tokens | $0.70 | $0.15 |
| Output price per 1M tokens | $2.10 | $0.50 |
| Context window | 262,144 | 262,144 |
| Image input | Yes | Yes |
| Developer | Mistral AI | NVIDIA |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, Mistral Large 4 (high), Terminus-2 or Nemotron 3.5 Super VL 120B (Reasoning), Terminus-2?
Mistral Large 4 (high), Terminus-2, 59 to 52.
Which is cheaper per task, Mistral Large 4 (high), Terminus-2 or Nemotron 3.5 Super VL 120B (Reasoning), Terminus-2?
Nemotron 3.5 Super VL 120B (Reasoning), Terminus-2, $0.0096 to $0.10.
Which writes fewer output tokens per task, Mistral Large 4 (high), Terminus-2 or Nemotron 3.5 Super VL 120B (Reasoning), Terminus-2?
Nemotron 3.5 Super VL 120B (Reasoning), Terminus-2, 7k to 14k.
Which has the larger context window, Mistral Large 4 (high), Terminus-2 or Nemotron 3.5 Super VL 120B (Reasoning), Terminus-2?
They are level, at 262,144 tokens.