Inference
EN
MiniMax

MiniMax M3.1 Flash

MiniMax M3.1 Flash is a reasoning model from MiniMax. 22 benchmarks count toward its index, in 5 categories.

  • Open weights
  • Reasoning
  • Input: text · image · video
  • Released 2026-10
RankedEnough results for an index.
IndexAll categories, out of 100.69.3±5.4Median 51.0
CoverageIndex weight that has results.80%
SpeedOutput tokens per second.100/sMedian 67/s
First tokenSeconds to the first output token.1.4 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$0.15Median $0.60
Output / 1MUS dollars per 1M output tokens.$0.60Median $2.55
ContextMaximum tokens in one request.1.05MMedian 262K
EloLMArena rating and rank.—Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
minimax/minimax-m3.1-flash
Max outputMaximum output tokens per request.
524K
Knowledge cutoffLast date of training data.
—
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.015
WeightsLink to the published weights.
Hugging Face (opens in a new tab) ↗
RetiresDate OpenRouter removes the model.
—

Run locally

220.49B parameters
16-bit
464.0 GB
8-bit
247.0 GB
4-bit
141.4 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy needs more than 128 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.10 counted71.818 / 218
CodingCode writing and repair.5 counted69.422 / 209
ReasoningLogic problems and puzzles.3 counted67.933 / 227
MultimodalTasks with images and text.2 counted58.146 / 84
KnowledgeFacts and expert knowledge.2 counted68.925 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.None counted——
MathMath problems.None counted——

Results

22 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
AutomationBenchAgenticAbove fitFar above this model's index.62.2%1 / 32
Artificial Analysis Humanity's Last ExamKnowledge48.0%17 / 222
LiveBench Browser UseAgentic92.0%4 / 49
NL2RepoCoding61.2%4 / 36
Critical Physics TasksReasoning22.3%25 / 214
Artificial Analysis Long Context ReasoningReasoning79.7%45 / 217
DeepSWEAgentic73.7%13 / 59
Artificial Analysis Omniscience AccuracyKnowledge45.1%37 / 164
GDPval-AA normalizedAgentic47.9%31 / 133
Artificial Analysis MMMU-ProMultimodal79.1%29 / 113
OSWorld 2.0Agentic61.7%7 / 25
PostTrainBench v1.1Coding39.3%5 / 15
Artificial Analysis SciCodeCoding53.4%39 / 114
Vibe Code Bench v1.1Coding73.9% ±4.245 / 123
Humanity's Last Exam without toolsKnowledge43.4%20 / 53
Terminal-Bench 4.0Agentic25.3% ±1.028 / 61
CharXiv ReasoningMultimodal82.5%23 / 49
LiveBench Coding 2.0Coding83.9%26 / 52
CyberGymAgentic82.2%15 / 29
LiveBench ResearchAgentic72.7%29 / 52
LiveBench Long AgenticAgentic34.7%29 / 51
LiveBench DesignAgentic74.0%31 / 52
LiveBench ReasoningReasoning82.9%48 / 80
Agents' Last ExamAgentic29.6%13 / 21
Agent Arena task outcomeAgentic−0.734 / 54
LiveBench Data AnalysisReasoning71.4%52 / 80
Agent Arena steerabilityAgentic−2.636 / 54
Multimodal Knowledge Agent BenchmarkAgentic33.5%18 / 19
Tax Agent BenchFinance32.1% ±3.323 / 82
Legal Research BenchLegal34.1% ±3.350 / 90
SAGEEducation85.3% ±2.018 / 29
HLAB-Vals-1.1Legal7.5% ±2.324 / 35
BabyVision with PythonMultimodal91.0%—
BankerToolBenchAgentic76.2%—
LOCA-Bench 256KAgentic60.3%—
SWE Refactor Bench: Can Coding Agents Complete a Long-Horizon, Whole-Repository Stack Migration?Agentic23.0%—

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseLiveBenchNo licence statedVals AINo licence statedLMArenaCC BY 4.0

History

5 releases
Index72.7 → 69.3
23 Oct 202612 Nov 2026
Output / 1M$0.60 → $0.60
23 Oct 202612 Nov 2026

Questions

How much does MiniMax M3.1 Flash cost?
$0.15 per 1M input tokens, $0.60 per 1M output tokens.
How good is MiniMax M3.1 Flash?
Its index is 69.3 out of 100. That is place 23 of 283 ranked models.
What is MiniMax M3.1 Flash best at?
Agentic: 71.8 out of 100, place 18 of 218.
How fast is MiniMax M3.1 Flash?
100 output tokens per second, 1.4 s to the first token.
What is the context window of MiniMax M3.1 Flash?
1.05M tokens.
Is MiniMax M3.1 Flash open source?
Yes. Its weights are published.
When was MiniMax M3.1 Flash released?
October 2026.

Badge

MiniMax M3.1 Flash on Inference360: place 23 of 283, index 69.3
HTML
<a href="https://inference360.ai/models/minimax-m3-1-flash"><img src="https://inference360.ai/badges/minimax-m3-1-flash.svg" alt="MiniMax M3.1 Flash on Inference360: place 23 of 283, index 69.3" height="20"></a>
Markdown
[![MiniMax M3.1 Flash on Inference360: place 23 of 283, index 69.3](https://inference360.ai/badges/minimax-m3-1-flash.svg)](https://inference360.ai/models/minimax-m3-1-flash)

Compare with

Grok 4.769.1GPT-5.569.1GPT-5.6 Terra68.9DeepSeek V4.1 Flash 101768.5Gemini 3.7 Flash68.5

Lower price

Ling 3.1 Flash67.2 · $0.135/1MDeepSeek V4.1 Flash 101768.5 · $0.40/1M

More from MiniMax

All models
MiniMax M3.1 Flash PreviewUnrankedMiniMax M355.7MiniMax M2.751.2MiniMax M2.550.2MiniMax M2-herUnranked