MiniMax-M3.1-Flash Dify and LangChain Integration Configuration

One-sentence takeaway

The same official MiniMax integration page provides Minimax plugin settings for Dify and langchain-openai OpenAI-compatible adapter settings for LangChain. Both use the exact model ID MiniMax-M3.1-Flash, but their API Base values are different: an Anthropic-compatible endpoint for Dify and an OpenAI-compatible endpoint for LangChain.

Use cases

• Suitable tasks: Building workflows, RAG, or Agents in Dify; calling MiniMax models from LangChain and integrating them into application pipelines.

• Unsuitable tasks: Complete production configurations that require reasoning, sampling, timeout, rate-limit, or pricing parameters not published on this page.

• Supported model version: MiniMax-M3.1-Flash.

• Supported client, Agent, or API: Dify Cloud; LangChain's langchain-openai / OpenAI-compatible interface.

• Recommended reasoning tier and parameters: Not specified. The source only provides the API Base, Subscription Key, and model ID; it does not publish reasoning_effort, temperature, or top_p.

Ready-to-use content

Minimum Dify configuration

Minimum LangChain configuration

First install the OpenAI-compatible adapter:

pip install langchain-openai

Replace sk-cp-... with your own MiniMax Subscription Key:

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="MiniMax-M3.1-Flash",
    api_key="sk-cp-...",
    base_url="https://api.minimax.io/v1",
)
print(llm.invoke("Hello").content)

Testing or workflow steps

Dify

1. Open Dify Cloud and sign in.

2. Go to Settings → Workspace → Model Provider, find Minimax, and click Install.

3. Click Add API Key, enter the Subscription Key, set API Base to https://api.minimax.io/anthropic, and leave Group ID blank.

4. Save the settings and use MiniMax-M3.1-Flash in a workflow.

LangChain

1. Install langchain-openai.

2. Create ChatOpenAI, fix model="MiniMax-M3.1-Flash", and set base_url="https://api.minimax.io/v1".

3. Pass the Subscription Key to api_key and call llm.invoke(...) to verify the returned content.

Original evidence and data

• The source page is titled Other Tools - MiniMax API Docs, and the exact model ID listed on the page is MiniMax-M3.1-Flash.

• The Dify section describes Dify as an open-source LLM application development platform that supports workflows, RAG, Agents, and observability. It also provides steps for plugin installation, API Key, API Base, and Group ID configuration.

• Dify's API Base is explicitly https://api.minimax.io/anthropic, and Group ID is explicitly required to be left blank.

• The LangChain section explicitly requires installing langchain-openai and uses ChatOpenAI, https://api.minimax.io/v1, and MiniMax-M3.1-Flash.

• The page's shared configuration reference also lists the OpenAI-compatible protocol Base URL https://api.minimax.io/v1 and the Anthropic-compatible protocol Base URL https://api.minimax.io/anthropic.

• The source does not provide a publication date, nor does it provide independent request results, performance data, quotas, rate limits, timeouts, or pricing information for Dify or LangChain.

Scope and limitations

• The Subscription Key is a credential placeholder; this article contains no real key. The sk-cp-... example must be replaced with the user's own key.

• The Dify configuration uses the Anthropic-compatible Base provided on the page, while the LangChain example uses the OpenAI-compatible Base provided on the page. Do not swap the two endpoints and treat the result as evidence from this article.

• The source only says that Dify can use the model in a workflow after the settings are saved. It does not provide plugin version, model dropdown behavior, call parameters, or error-handling details.

• The source gives only one LangChain invoke("Hello") example. It cannot support conclusions about tool calling, streaming output, multimodal input, or long-running compatibility.

• The page does not publish the reasoning tier, sampling parameters, timeout, retry, rate-limit, pricing, context window, or maximum output limits. These fields remain unspecified.

• This article does not contain a complete model prompt. It is categorized as configuration and does not claim that a copyable model-specific prompt exists.

Reproduction notes

On 2026-10-21, the official page was opened and read through a Tabbit exclusive task, M31 prompt11 apps. The two Dify and LangChain sections on the same page, the exact model ID, endpoints, fields, and steps were checked. The original extracted evidence was saved to /tmp/m31-apps-evidence.txt. Dify was not signed in to, LangChain was not installed, no model request was sent, and no key was configured or saved.