Image creation
MiniMax M3.1 Flash - MiniMax 3.1 Flash
MiniMax’s fast, low-cost model for coding, agents, and computer use, with a 1M-token context.
Product insights
MiniMax M3.1 Flash at a glance
Value proposition
Near-frontier agent and coding performance at Flash-tier speed and price.
Problem solved
Agent loops burn through millions of tokens, and top models make that slow and expensive.
Audience
Developers building coding agents, automation, and computer-use tools on a budget.
Pricing and access
API: $0.15 per 1M input tokens, $0.60 per 1M output tokens, $0.015 per 1M cached input tokens. Also included in the M Plan subscription and MiniMax Code.
Market
- Competes with low-cost fast models such as DeepSeek V4.1 Flash and Gemini Flash.
- A larger M3.1 Pro model is announced as coming later.
Tech stack
- 1M-token context (input + output combined)
- Text, image, and video input; text output
- OpenAI- and Anthropic-compatible API endpoints
01Why it is here
What is MiniMax M3.1 Flash?
MiniMax M3.1 Flash is the fast, low-cost member of MiniMax’s M3.1 model family. MiniMax reports scores of 73.7% on DeepSWE v1.1, 61.7 on OSWorld V2, 82% on CyberGym, and 76.2 on BankerToolBench, and Artificial Analysis ranks it near the top of its price class. These are mostly vendor-reported numbers, so test it on your own tasks. Thinking is always on, which helps quality but adds latency and output tokens.
02Workflow
How to use MiniMax M3.1 Flash
- 01
Try it in MiniMax Code or the agent app
Use the MiniMax Code desktop app or MiniMax Agent on the web to try M3.1 Flash on real coding and agent tasks.
- 02
Get an API key
Sign up on the MiniMax developer platform and get a key, or pick it on OpenRouter.
- 03
Plug it into your tools
Point any OpenAI- or Anthropic-compatible client, such as Claude Code or Cline, at the MiniMax endpoint and select MiniMax-M3.1-Flash.
03Capabilities
Key features of MiniMax M3.1 Flash
- Built for coding, agents, and computer use
- 1M-token context window
- 170+ tokens/sec typical output speed (vendor-reported)
- $0.15 / $0.60 per 1M input / output tokens
- Accepts image and video input
- Available via API, M Plan, MiniMax Code, and OpenRouter
04Use cases
What is MiniMax M3.1 Flash best for?
- Running coding agents at low cost
- High-volume automation and tool-calling workflows
- Long-context work over large codebases or documents
05Questions
FAQ about MiniMax M3.1 Flash
What is MiniMax 3.1 Flash?
It is MiniMax M3.1 Flash, a fast, low-cost language model from MiniMax released on October 19, 2026, aimed at coding, AI agents, and computer use.
How much does MiniMax M3.1 Flash cost?
API pricing is $0.15 per million input tokens and $0.60 per million output tokens, with cached input at $0.015 per million.
Is MiniMax M3.1 Flash the Space Bunny model?
Yes. MiniMax confirmed that Space Bunny, the anonymous stealth model that led OpenRouter’s usage charts, was an earlier version of M3.1 Flash.
Is MiniMax M3.1 Flash open source?
Not yet. At launch it is available through the API and MiniMax’s apps, and MiniMax says the open weights will be released later.
Official website snapshot
minimax.io
MiniMax M3.1 Flash (often searched as MiniMax 3.1 Flash) is a language model from Chinese AI company MiniMax, launched on October 19, 2026. It is built for agentic work: coding, tool use, computer use, and cybersecurity tasks. It takes text, image, and video input, returns text, and has a 1M-token context window. MiniMax says it runs about 3x faster than M3 at roughly a third of the cost, with typical output above 170 tokens per second. API pricing is $0.15 per million input tokens and $0.60 per million output tokens. It was tested anonymously as the "Space Bunny" stealth model, which topped OpenRouter’s usage charts before the reveal.
