
MiniMax-M3
MiniMax-M3 is an open-weight model from MiniMax, released May 31, 2026. It ranks #93 of 351 on the LMArena text leaderboard with a rating of 1432. It costs $0.30 per million input tokens and $1.20 per million output tokens, with a 1M-token context window.
- Developer
- MiniMax
- Released
- May 31, 2026
- Context window
- 1M
- Price in / out, per 1M tokens
- $0.30 / $1.20
- Weights
- Open · MiniMax Community License
- Inputs
- Text, Images, Video
- Reasoning
- Yes
- Tool use
- Yes
- Hugging Face downloads, 30 days
- 203.5K
- Hugging Face likes
- 1.6K
The rating LMArena published at each update since Jun 3, 2026. Ratings move as votes come in and new models join.
Benchmarks
| Benchmark | Score bar | Accuracy | Rank | Cost / test |
|---|---|---|---|---|
| Terminal-Bench 4.0Coding | 1.01% | 51 / 55 | $7.23 | |
| SWE-benchCoding | 75.00% | 43 / 85 | $0.42 | |
| Vibe Code BenchCoding | 47.57% | 64 / 115 | $6.45 | |
| Code MigrationCoding | 20.90% | 26 / 28 | $8.39 | |
| LiveCodeBenchCoding | 82.15% | 56 / 143 | $0.6 / $2.4 | |
| Finance AgentFinance | 54.34% | 28 / 31 | $0.30 | |
| Finance IndexFinance | 48.37% | 27 / 31 | $1.22 | |
| Tax Agent BenchFinance | 20.59% | 50 / 76 | $0.16 | |
| Legal Research BenchLaw | 29.81% | 52 / 85 | $0.34 | |
| MedCodeMedicine | 46.29% | 37 / 107 | $0.6 / $2.4 | |
| GPQA DiamondReasoning | 92.68% | 15 / 138 | $0.6 / $2.4 | |
| MMLU-ProReasoning | 84.22% | 66 / 138 | $0.6 / $2.4 | |
| MMMUReasoning | 81.16% | 41 / 93 | $0.6 / $2.4 |
Vals AI, as of Oct 24, 2026.
LMArena, by category
| Leaderboard | Rating | Rank | Votes |
|---|---|---|---|
| TextLMArena | 1432 | 93 / 351 | 67K |
| CodingLMArena | 1470 | 69 / 346 | 19K |
| MathLMArena | 1429 | 106 / 339 | 3.4K |
| Creative writingLMArena | 1403 | 94 / 349 | 12.4K |
| Instruction followingLMArena | 1432 | 71 / 351 | 23.3K |
| Hard promptsLMArena | 1447 | 82 / 351 | 44.3K |
| Multi-turnLMArena | 1443 | 83 / 349 | 11.6K |
| Long queriesLMArena | 1445 | 71 / 330 | 30.2K |
| ExpertLMArena | 1461 | 72 / 305 | 7.3K |
| Legal and governmentIndustries | 1436 | 100 / 325 | 5.4K |
| Medicine and healthcareIndustries | 1439 | 99 / 325 | 4.5K |
| Business and financeIndustries | 1434 | 70 / 344 | 12.9K |
| ChineseLanguages | 1465 | 95 / 328 | 4.3K |
| SpanishLanguages | 1432 | 98 / 249 | 2.1K |
| GermanLanguages | 1430 | 84 / 261 | 1.1K |
| FrenchLanguages | 1443 | 101 / 252 | 2.5K |
| JapaneseLanguages | 1379 | 100 / 236 | 1K |
| WebDevLMArena | 1482 | 67 / 124 | 20.3K |
| VisionLMArena | 1251 | 62 / 146 | 19.8K |
LMArena, CC BY 4.0, as of Oct 22, 2026. Where a model is rated at several effort levels, its best is shown.
Compare MiniMax-M3
MiniMax
Builds long-context text, voice and video models, with open-weight releases. Founded 2021, based in Shanghai. minimax.io
Questions
Who makes MiniMax-M3?
MiniMax-M3 is made by MiniMax, founded 2021 and based in Shanghai.
How much does MiniMax-M3 cost?
$0.30 per million input tokens and $1.20 per million output tokens, at its provider’s list price.
What is MiniMax-M3’s context window?
1M tokens (1,048,576).
How does MiniMax-M3 rank?
#93 of 351 on LMArena’s text leaderboard, rated 1432 (95% interval 1428–1436) from 67,030 votes, as of Oct 22, 2026.
Is MiniMax-M3 open source?
MiniMax-M3 has open weights under the MiniMax Community License license, downloadable from Hugging Face (MiniMaxAI/Minimax-M3).
Gemini 4 Argon
Claude Opus 5.5
Muse Wonder
MiMo-V2.6-Pro
Qwen3.8-Max
Kimi K3
GLM-5.3
GPT-5.5
ERNIE 5.1
DeepSeek-V4.1-Flash-20260910
Grok 4.20 Beta
amazon-nova-experimental-chat
Dola Seed 2.0 Pro
NVIDIA-Nemotron-3-Ultra-550B-A55B
Gemma 4 31B
Step 5 Preview