MiniMax/MiniMax-M3.1-Flash Model Info-阿里云帮助中心

MiniMax M3.1 Flash is a high-performance multimodal model for agentic tasks. It supports a 1-million-token context window and balances quality, inference speed, and cost in complex tasks, software development, and cybersecurity scenarios.

Inference Service Provider

The inference service provider for MiniMax/MiniMax-M3.1-Flash is MiniMax.

Model Capabilities

Capability	Support
Input Modality	Image Text Video
Output Modality	Text
Deep Thinking	Supported and always enabled
Function Calling	Supported
Context Caching	Supported

Context Limits

Parameter	Value
Context Window	1,000,000 tokens

Pricing

This page only shows the original pricing for model API calls, excluding any limited-time promotions. Visit Model Studio Console for promotional offers.

China (Beijing)

Billing Item	Price (CNY)	Unit
Input	1	Per 1M tokens
Output	4	Per 1M tokens
Input (Implicit Cache)	0.1	Per 1M tokens

Rate Limits

China (Beijing)

Parameter	Value
RPM (Requests Per Minute)	200
TPM (Tokens Per Minute)	10,000,000

The service may also enforce per-second limits based on RPS (RPM/60) and TPS (TPM/60). For more information, see Rate limits.

Usage limitations

stop is not supported. Passing it does not truncate the output at the specified stop sequence. The request may still succeed and return finish_reason: "stop", but this does not mean that the specified stop sequence took effect.
The model does not support using parallel_tool_calls to control whether tool calls are parallel. Setting it to false does not guarantee sequential tool calls, and the model may still return multiple tool_calls in one turn.
tool_choice supports only auto. Other values may return an error or have no effect. For example, specifying a named function does not guarantee that the model calls it.

Invocation

Call the model using the OpenAI-compatible API. For instructions and parameter differences from MiniMax-M3, see MiniMax.