MiniMax M3
MiniMax released 2026-05-31 mit
DeepSeek V4 Flash 0731 scores higher and costs less — but you would give something up.
+6.4 on the capability index and 80% cheaper ($0.105/M against $0.525/M). What you lose:
- 512K → 384K max output
- no image, video input
Capability scores do not measure context length, output ceiling or which inputs a model accepts, so a higher score does not mean a drop-in replacement.
Every price dimension
| Input | $0.300/M |
|---|---|
| Output | $1.20/M |
| Cached input | $0.060/M |
Independent scores
| Intelligence | 45.4 |
|---|---|
| Coding | 58.6 |
| Agentic | 36.1 |
| LMArena Elo | 1434.8default |
Sources are listed separately rather than averaged. Across the 62 models both have scored they correlate at r = 0.806 — close agreement overall, but four models rank very differently between them, and a blended score would hide exactly those.
Best rankings by task
- python-pptxslides #10of 35 1241
- agenticgamedev #13of 34 1179
- htmlslides #13of 34 1177
- mobileapps #14of 56 1214
- fullstack #16of 59 1232
- webapps #18of 56 1237
- androidnative #22of 54 1179
- website #25of 151 1279
Capability
| Context window | 1.0M |
|---|---|
| Max output | 512K |
| Input modes | text, image, video |
| Tool use | yes |
| Reasoning | optional |
| Open weights | yes |
Provenance
| Price source | openrouter.ai |
|---|---|
| Fetched | 2026-08-24 |
| Quality data | verified |
| Cross-checked | vendor page |
MiniMax-M3 is a multimodal foundation model from MiniMax. It supports text, image, and video inputs with text output, a 1M-token context window, and is suited for long-horizon agentic work, coding,...