GLM-5.3-Flash (low) vs Mistral Large 4 (high)
Results updated 3 November 2026 · Both in Terminus-2 · How we benchmark
Verdict
GLM-5.3-Flash (low) beats Mistral Large 4 (high) by 3 points, and is cheaper and faster.
- Cost per task
- $0.048 vs $0.10
- Time per task
- 4.2 min vs 12.0 min
- Tokens per attempt
- 476k vs 226k
- Image input
- Yes vs Yes
GLM-5.3-Flash (low), Terminus-2 scores higher on Earth-Bench V2 and GeoBenchX; Mistral Large 4 (high), Terminus-2 scores higher on GS-QA2. GLM-5.3-Flash (low), Terminus-2 is cheaper and faster: $0.048 against $0.10 per task, and 4.2 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 | GLM-5.3-Flash (low) | Mistral Large 4 (high) |
|---|---|---|
| Recognising infeasible requests | 47 | 23 |
| Climate and time series | 59 | 40 |
| Making maps | 67 | 82 |
| Spatial statistics | 38 | 46 |
Where the tasks are set: each kind of work links to its tasks and scoresCharts for each: GLM-5.3-Flash (low), Terminus-2 · Mistral Large 4 (high), Terminus-2.
Side by side
| Metric | GLM-5.3-Flash (low), Terminus-2 | Mistral Large 4 (high), Terminus-2 |
|---|---|---|
| Geospatial Agent Index: Text only | 62 | 59 |
| Geospatial Agent Index: Text + vision | No data | No data |
| Earth-Bench V2 score | 53 | 46 |
| GeoBenchX score | 57 | 55 |
| GS-QA2 score | 76 | 77 |
| Vision agents score | No data | No data |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 94 | 80 |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 95 | 94 |
| OGD4All score (standalone; in no index) | No data | 96 |
| GeoNatureAgent score (standalone; in no index) | No data | 100 |
| MultiGlobeQA score (standalone; in no index) | No data | 62 |
| GeoAgentEval score (standalone; in no index) | No data | 60 |
| Remote sensing and raster analysis score | 67 | 67 |
| GIS analysis score | 75 | 82 |
| Climate and time series score | 59 | 40 |
| Spatial statistics score | 38 | 46 |
| Making maps score | 67 | 82 |
| Recognising infeasible requests score | 47 | 23 |
| Cost per task | $0.048 | $0.10 |
| Time per task (at 100 output tokens/s) | 4.2 min | 12.0 min |
| Tokens per attempt | 476k | 226k |
| Turns per attempt | 14.0 | 13.8 |
| Input price per 1M tokens | $0.15 | $0.70 |
| Output price per 1M tokens | $0.50 | $2.10 |
| Context window | 1,048,576 | 262,144 |
| Image input | Yes | Yes |
| Developer | Z.ai | Mistral AI |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, GLM-5.3-Flash (low), Terminus-2 or Mistral Large 4 (high), Terminus-2?
GLM-5.3-Flash (low), Terminus-2, 62 to 59.
Which is cheaper per task, GLM-5.3-Flash (low), Terminus-2 or Mistral Large 4 (high), Terminus-2?
GLM-5.3-Flash (low), Terminus-2, $0.048 to $0.10.
Which writes fewer output tokens per task, GLM-5.3-Flash (low), Terminus-2 or Mistral Large 4 (high), Terminus-2?
Mistral Large 4 (high), Terminus-2, 14k to 15k.
Which has the larger context window, GLM-5.3-Flash (low), Terminus-2 or Mistral Large 4 (high), Terminus-2?
GLM-5.3-Flash (low), Terminus-2, 1,048,576 tokens to 262,144 tokens.