Mistral Large 4 (high) vs Qwen 3.6 35B-A3B (Reasoning)
Results updated 6 November 2026 · Both in Terminus-2 · How we benchmark
Verdict
Mistral Large 4 (high) beats Qwen 3.6 35B-A3B (Reasoning) by 3 points; Qwen 3.6 35B-A3B (Reasoning) is cheaper and faster.
- Cost per task
- $0.10 vs $0.068
- Time per task
- 11.4 min vs 6.2 min
- Tokens per attempt
- 233k vs 358k
- Image input
- Yes vs Yes
Mistral Large 4 (high), Terminus-2 scores higher on GeoBenchX and GS-QA2; Qwen 3.6 35B-A3B (Reasoning), Terminus-2 scores higher on Earth-Bench V2. Qwen 3.6 35B-A3B (Reasoning), Terminus-2 is cheaper and faster: $0.068 against $0.10 per task, and 6.2 min against 11.4 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) | Qwen 3.6 35B-A3B (Reasoning) |
|---|---|---|
| Climate and time series | 39 | 62 |
| GIS analysis | 85 | 68 |
| Making maps | 84 | 70 |
| Remote sensing and raster analysis | 70 | 67 |
Where the tasks are set: each kind of work links to its tasks and scoresCharts for each: Mistral Large 4 (high), Terminus-2 · Qwen 3.6 35B-A3B (Reasoning), Terminus-2.
Side by side
| Metric | Mistral Large 4 (high), Terminus-2 | Qwen 3.6 35B-A3B (Reasoning), Terminus-2 |
|---|---|---|
| Geospatial Agent Index: Text only | 63 | 60 |
| Geospatial Agent Index: Text + vision | No data | No data |
| Earth-Bench V2 score | 50 | 62 |
| GeoBenchX score | 59 | 51 |
| GS-QA2 score | 79 | 69 |
| Vision agents score | No data | No data |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 80 | 65 |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 94 | 58 |
| OGD4All score (standalone; in no index) | 95 | 93 |
| GeoNatureAgent score (standalone; in no index) | 100 | 93 |
| MultiGlobeQA score (standalone; in no index) | 71 | 72 |
| GeoAgentEval score (standalone; in no index) | 92 | 84 |
| ImpactMesh score (standalone; in no index) | No data | No data |
| Remote sensing and raster analysis score | 70 | 67 |
| Change detection score | 62 | 60 |
| GIS analysis score | 85 | 68 |
| Climate and time series score | 39 | 62 |
| Making maps score | 84 | 70 |
| Cost per task | $0.10 | $0.068 |
| Time per task (at 100 output tokens/s) | 11.4 min | 6.2 min |
| Tokens per attempt | 233k | 358k |
| Turns per attempt | 14.3 | 16.0 |
| Input price per 1M tokens | $0.70 | $0.25 |
| Output price per 1M tokens | $2.10 | $1.49 |
| Context window | 262,144 | 262,144 |
| Image input | Yes | Yes |
| Developer | Mistral AI | Alibaba (Qwen) |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, Mistral Large 4 (high), Terminus-2 or Qwen 3.6 35B-A3B (Reasoning), Terminus-2?
Mistral Large 4 (high), Terminus-2, 63 to 60.
Which is cheaper per task, Mistral Large 4 (high), Terminus-2 or Qwen 3.6 35B-A3B (Reasoning), Terminus-2?
Qwen 3.6 35B-A3B (Reasoning), Terminus-2, $0.068 to $0.10.
Which writes fewer output tokens per task, Mistral Large 4 (high), Terminus-2 or Qwen 3.6 35B-A3B (Reasoning), Terminus-2?
Mistral Large 4 (high), Terminus-2, 14k to 16k.
Which has the larger context window, Mistral Large 4 (high), Terminus-2 or Qwen 3.6 35B-A3B (Reasoning), Terminus-2?
They are level, at 262,144 tokens.