The official Cursor integration page requires clearing OPENAI_API_KEY and OPENAI_BASE_URL first, then enabling the global OpenAI Base URL override under Cursor's Models → API Keys, entering https://api.minimax.io/v1 and the MiniMax Subscription Key, and finally adding and enabling the exact model name MiniMax-M3.1-Flash; custom models apply only to Chat, Composer, and Edit, not Cursor Tab autocomplete.
Suitable tasks: Using MiniMax-M3.1-Flash in Cursor for Chat, Composer, or Edit coding tasks.
Unsuitable tasks: Tasks that depend on Cursor Tab autocomplete, or scenarios that require different Base URLs for different models at the same time.
Applicable model version: MiniMax-M3.1-Flash.
Applicable client, Agent, or API: Cursor; the MiniMax OpenAI-compatible API.
Recommended reasoning tier and parameters: Not specified. The page exposes only the Base URL, authentication fields, and model name.
Clear potentially conflicting environment variables; this article records the official requirements only and has not run them locally:
unset OPENAI_API_KEY
unset OPENAI_BASE_URLInstall Cursor, open it, click Log In, and sign in to your Cursor account.
Open Models on the left and expand API Keys.
Turn on Override OpenAI Base URL.
Enter https://api.minimax.io/v1 as the Base URL.
Paste the MiniMax Subscription Key into the OpenAI API Key field.
Turn on Use OpenAI API Key, then click Enable OpenAI API Key in the confirmation dialog.
At the bottom of the model list, click Add Custom Model, enter the exact name MiniMax-M3.1-Flash, and click Add.
Enable the new model, select MiniMax-M3.1-Flash in the chat panel, and start developing.
Confirm that the Cursor account has Cursor Pro or a higher plan; otherwise, the page gives this error: The model MiniMax-M3.1-Flash does not work with your current plan or api key.
Complete the API configuration and add the custom model using the path above.
Select MiniMax-M3.1-Flash in Chat, Composer, or Edit, send a short coding task containing no sensitive information, and confirm that the model returns content.
If no content is returned, try setting HTTP Compatibility Mode to HTTP/1.0 under Browser & Network in Cursor settings, then retry.
The page title is Cursor, and it says that its purpose is to bring the latest MiniMax M-series models into Cursor.
The official requirements are to clear OPENAI_API_KEY and OPENAI_BASE_URL to avoid conflicts with the MiniMax API.
The Base URL is https://api.minimax.io/v1; the authentication value is the MiniMax Subscription Key, which is pasted into Cursor's OpenAI API Key field.
The exact custom model name is MiniMax-M3.1-Flash. The page specifically requires preserving its capitalization and hyphenation.
The page requires Cursor Pro or a higher plan to support custom models.
Override OpenAI Base URL is a global setting that affects all API keys configured in Cursor, including built-in Anthropic / GPT models. The page says that Cursor currently does not support setting different Base URLs per model.
Cursor Tab autocomplete does not use the custom MiniMax key; Tab uses Cursor's own model. Custom models apply only to Chat, Composer, and Edit.
The page does not disclose reasoning tiers, a thinking toggle, timeouts, quotas, or other parameter limits.
This article retains the https://api.minimax.io/v1 specified in the page body. The page also notes that screenshots use an endpoint for mainland China; do not mix the address in the screenshot with the body configuration, and do not infer endpoints for other regions from it.
Because the Base URL override is global, enabling it may affect Cursor's native Claude / GPT models; the page recommends turning off the override when MiniMax is not in use.
Cursor controls the model and billing for Cursor Tab; the custom model configuration for Chat / Composer / Edit cannot be extrapolated to Tab autocomplete.
The Subscription Key, Cursor Pro eligibility, and account availability depend on the corresponding accounts; this article contains no key and does not actually log in or write to the configuration.
The page does not provide a complete model prompt, so this is classified as configuration and does not claim that a copyable prompt exists.
On 2026-10-21, the official page was opened and read through the Tabbit isolated task M31 prompt07 cursor, checking the model name, client fields, endpoint, configuration path, Cursor Pro requirement, global Base URL limitation, available modes, and HTTP/1.0 compatibility mode. The evidence is saved at /tmp/m31-cursor-evidence.txt. No installation, login, request sending, or local configuration was performed.
MiniMax M3.1 Flash