Mistral Large 4 (high) vs GPT-6.1 Sol (xhigh)
Results updated 3 November 2026 · In Terminus-2 and Codex, so part of any gap can come from the harness · How we benchmark
Verdict
GPT-6.1 Sol (xhigh) clearly beats Mistral Large 4 (high) by 23 points, and is cheaper and faster.
- Cost per task
- $0.10 vs $0.085
- Time per task
- 12.0 min vs 0.8 min
- Tokens per attempt
- 226k vs 145k
- Image input
- Yes vs Yes
GPT-6.1 Sol (xhigh), Codex scores higher than Mistral Large 4 (high), Terminus-2 on all 3 evaluations scored for both. GPT-6.1 Sol (xhigh), Codex is cheaper and faster: $0.085 against $0.10 per task, and 0.8 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) | GPT-6.1 Sol (xhigh) |
|---|---|---|
| Recognising infeasible requests | 23 | 87 |
| Climate and time series | 40 | 75 |
| Remote sensing and raster analysis | 67 | 85 |
| GIS analysis | 82 | 93 |
Where the tasks are set: each kind of work links to its tasks and scoresCharts for each: Mistral Large 4 (high), Terminus-2 · GPT-6.1 Sol (xhigh), Codex.
Side by side
| Metric | Mistral Large 4 (high), Terminus-2 | GPT-6.1 Sol (xhigh), Codex |
|---|---|---|
| Geospatial Agent Index: Text only | 59 | 82 |
| Geospatial Agent Index: Text + vision | No data | 84 |
| Earth-Bench V2 score | 46 | 72 |
| GeoBenchX score | 55 | 79 |
| GS-QA2 score | 77 | 94 |
| Vision agents score | No data | 90 |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 80 | 100 |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 94 | 95 |
| OGD4All score (standalone; in no index) | 96 | No data |
| GeoNatureAgent score (standalone; in no index) | 100 | No data |
| MultiGlobeQA score (standalone; in no index) | 62 | No data |
| GeoAgentEval score (standalone; in no index) | 60 | No data |
| Remote sensing and raster analysis score | 67 | 85 |
| GIS analysis score | 82 | 93 |
| Climate and time series score | 40 | 75 |
| Spatial statistics score | 46 | 38 |
| Making maps score | 82 | 84 |
| Recognising infeasible requests score | 23 | 87 |
| Cost per task | $0.10 | $0.085 |
| Time per task (at 100 output tokens/s) | 12.0 min | 0.8 min |
| Tokens per attempt | 226k | 145k |
| Turns per attempt | 13.8 | 7.4 |
| Input price per 1M tokens | $0.70 | $2.00 |
| Output price per 1M tokens | $2.10 | $10.00 |
| Context window | 262,144 | 1,050,000 |
| Image input | Yes | Yes |
| Developer | Mistral AI | OpenAI |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, Mistral Large 4 (high), Terminus-2 or GPT-6.1 Sol (xhigh), Codex?
GPT-6.1 Sol (xhigh), Codex, 82 to 59.
Which is cheaper per task, Mistral Large 4 (high), Terminus-2 or GPT-6.1 Sol (xhigh), Codex?
GPT-6.1 Sol (xhigh), Codex, $0.085 to $0.10.
Which writes fewer output tokens per task, Mistral Large 4 (high), Terminus-2 or GPT-6.1 Sol (xhigh), Codex?
GPT-6.1 Sol (xhigh), Codex, 3k to 14k.
Which has the larger context window, Mistral Large 4 (high), Terminus-2 or GPT-6.1 Sol (xhigh), Codex?
GPT-6.1 Sol (xhigh), Codex, 1,050,000 tokens to 262,144 tokens.