None
None
Long Story Video Skill
Turn your idea into 10s–10min video—AI writes scripts, prompts, and generates footage automatically.
Ads Video Skill
Generate professional ads and sales videos—AI auto-generates scripts, prompts, and footage.
3D Science Explainer Video Skill
Convert scientific concepts into stunning 3D explain animations
Feedback
freeTrialImage.bannerPity
freeTrialImage.upgradeUnlock
- ✓freeTrialImage.benefitHd
- ✓freeTrialImage.benefitWatermark
- ✓freeTrialImage.benefitUnlimited
minimax m3.1 flash
Put MiniMax M3.1 Flash to work on agents, code and computer-use tasks — blazing 200+ TPS output with flat, low-cost tokens.
All Tools
Discover our comprehensive AI-powered animation toolkit
MiniMax H3
MiniMax H3 AI Video Generator
Seedance 2.5
The Future of AI Video Is Here.
Seedance 2.0
The Future of AI Video Is Here.
Veo3.1
Create Stunning Videos with Veo3.1
FLUX 3 Video Generator
Kling 3.0
Next-Gen AI Video Generator
Grok Video Generator
Create Videos from Text or Images with AI
MiniMax H3 video generator
MiniMax H3 AI Video Generator
MiniMax M3.1 Flash: A Fast Engine for Real Agent Work
Succeeding the anonymously operated Space Bunny release, MiniMax M3.1 Flash has been measured on enterprise automation, software engineering, computer operation and security tasks.
- Multi-Step Agent AutomationPlaces near the top of AutomationBench-AA and posts 76.2 on BankerToolBench, carrying SaaS workflows end to end across sales, marketing, operations, service, finance and HR.
- Software Engineering SupportScores 73.7% on DeepSWE v1.1 — reading repositories, shipping features, squashing defects and verifying its own changes with tests across long step chains.
- Desktop Operation and SecurityRecords 61.7 on OSWorld V2 and 29.6 on Agents' Last Exam for real-environment control, plus 82% on CyberGym for discovering and validating vulnerabilities.
Running MiniMax M3.1 Flash in Three Moves
Go from your first call to production-grade agent work in three short steps with MiniMax M3.1 Flash.
Core Strengths of MiniMax M3.1 Flash
Quicker responses, predictable token rates and proven results spanning automation, coding, desktop control and security engineering.
Deeper Sparsity, Quicker Replies
A sparser serving stack pushes routine output past 200 TPS with P50 holding above 170 TPS — close to three times the speed of the earlier release.
Simple, Predictable Token Rates
$0.15 per 1M input, $0.60 per 1M output and $0.015 per 1M cached input bring agentic workloads roughly 70% below what the previous model cost.
SaaS Workflows Run End to End
A leading AutomationBench-AA position and 76.2 on BankerToolBench show multi-step planning, financial compliance and chained tool calls inside live SaaS systems.
Real Software Engineering Depth
73.7% on DeepSWE v1.1 places it alongside top models on work that requires independent code reading, feature delivery, bug fixing and test validation.
Control of Real Desktop Environments
61.7 on OSWorld V2 and 29.6 on Agents' Last Exam — best among Chinese models on both — reflect hands-on tool use and finished results in live environments.
Security Work at Low Cost
82% on CyberGym, confirmed by internal expert review, supports code auditing, vulnerability patching, penetration testing, malware analysis and reverse engineering.
MiniMax M3.1 Flash: Your Questions Answered
Answers on MiniMax M3.1 Flash pricing, output speed, benchmark scores and the ways to get access.
So what is MiniMax M3.1 Flash, exactly?
It is the next Flash-generation model from MiniMax, grown out of the anonymously run Space Bunny system that handled more than 200 trillion tokens; it moves ahead on agentic work, coding, computer use and cybersecurity.
What does a million tokens cost?
Pricing is flat: $0.15 for 1M input tokens, $0.60 for 1M output tokens and $0.015 for 1M cached input tokens, leaving overall agentic usage about 70% cheaper than the model it replaces.
Why is inference nearly three times faster?
Greater sparsity throughout the serving stack: everyday output reaches 200+ TPS while the median (P50) rate stays above 170 TPS.
Which benchmarks support these numbers?
A top-tier AutomationBench-AA placement, 76.2 on BankerToolBench, 73.7% on DeepSWE v1.1, 61.7 on OSWorld V2, 29.6 on Agents' Last Exam and 82% on CyberGym.
How do I get access to the model?
Use the MiniMax API at platform.minimax.io, subscribe to M Plan or install MiniMax Code — the desktop build downloads from agent.minimax.io/download; weights arrive later.
How does Flash differ from the Pro model?
Flash continues MiniMax's push to balance efficiency with capability, while Pro — built to extend the frontier of intelligence — lands at a later date.
Put MiniMax M3.1 Flash to Work Today
Fire up the MiniMax API, join M Plan or install MiniMax Code and get frontier agent, coding, computer-use and security performance at about 70% lower spend with near-3x faster inference.