MiniMax M3 (Reasoning) vs Claude Opus 5.5 (high)
Results updated 1 November 2026 · In MiniMax Code and Claude Code, so part of any gap can come from the harness · How we benchmark
Verdict
Claude Opus 5.5 (high) clearly beats MiniMax M3 (Reasoning) by 14 points, and is cheaper and faster.
- Cost per task
- $0.15 vs $0.13
- Time per task
- 4.9 min (estimated) vs 0.7 min
- Tokens per attempt
- 1.68M vs 108k
- Image input
- Yes vs Yes
Claude Opus 5.5 (high), Claude Code scores higher than MiniMax M3 (Reasoning), MiniMax Code on all 3 evaluations scored for both. Claude Opus 5.5 (high), Claude Code is cheaper and faster: $0.13 against $0.15 per task, and 0.7 min against 4.9 min.
Where they differ most
By kind of work, share of tasks passed, 0 to 100 · Higher is better
| Kind of work | MiniMax M3 (Reasoning) | Claude Opus 5.5 (high) |
|---|---|---|
| Recognising infeasible requests | 33 | 78 |
| Remote sensing and raster analysis | 73 | 88 |
| Making maps | 77 | 86 |
| GIS analysis | 81 | 89 |
Where the tasks are set: each kind of work links to its tasks and scoresCharts for each: MiniMax M3 (Reasoning), MiniMax Code · Claude Opus 5.5 (high), Claude Code.
Side by side
| Metric | MiniMax M3 (Reasoning), MiniMax Code | Claude Opus 5.5 (high), Claude Code |
|---|---|---|
| Geospatial Agent Index: Text only | 68 | 82 |
| Geospatial Agent Index: Text + vision | No data | 84 |
| Earth-Bench V2 score | 68 | 79 |
| GeoBenchX score | 59 | 79 |
| GS-QA2 score | 78 | 90 |
| Vision agents score | No data | 89 |
| GeoAgentBench score (Guided workflows; not in the Agent Index) | 94 | 100 |
| GeoAnalystBench score (Guided workflows; not in the Agent Index) | 79 | 100 |
| OGD4All score (standalone; in no index) | 94 | 95 |
| GeoNatureAgent score (standalone; in no index) | 93 | 100 |
| MultiGlobeQA score (standalone; in no index) | 73 | 78 |
| GeoAgentEval score (standalone; in no index) | 92 | 100 |
| Remote sensing and raster analysis score | 73 | 88 |
| GIS analysis score | 81 | 89 |
| Climate and time series score | 74 | 75 |
| Spatial statistics score | 62 | 54 |
| Making maps score | 77 | 86 |
| Recognising infeasible requests score | 33 | 78 |
| Cost per task | $0.15 | $0.13 |
| Time per task (at 100 output tokens/s) | 4.9 min (estimated) | 0.7 min |
| Tokens per attempt | 1.68M | 108k |
| Turns per attempt | No data | 5.3 |
| Input price per 1M tokens | $0.30 | $4.00 |
| Output price per 1M tokens | $1.20 | $20.00 |
| Context window | 1,048,576 | 1,000,000 |
| Image input | Yes | Yes |
| Developer | MiniMax | Anthropic |
Frequently asked questions
Which scores higher on the Geospatial Agent Index, MiniMax M3 (Reasoning), MiniMax Code or Claude Opus 5.5 (high), Claude Code?
Claude Opus 5.5 (high), Claude Code, 82 to 68.
Which is cheaper per task, MiniMax M3 (Reasoning), MiniMax Code or Claude Opus 5.5 (high), Claude Code?
Claude Opus 5.5 (high), Claude Code, $0.13 to $0.15.
Which writes fewer output tokens per task, MiniMax M3 (Reasoning), MiniMax Code or Claude Opus 5.5 (high), Claude Code?
Claude Opus 5.5 (high), Claude Code, 2k to 29k.
Which has the larger context window, MiniMax M3 (Reasoning), MiniMax Code or Claude Opus 5.5 (high), Claude Code?
MiniMax M3 (Reasoning), MiniMax Code, 1,048,576 tokens to 1,000,000 tokens.