canonical: https://jentic.com/apis/minimaxi.com/minimaxi

# Minimaxi MiniMax API

MiniMax API is a multimodal generation platform that exposes MiniMax's large language, text-to-speech, video, and music models behind a single REST interface. An agent can generate a chat completion, synthesize speech, produce a short video, or compose a music track, then upload and list the files those jobs consume. It suits teams building assistants or content pipelines that need several media types from one provider.

## For AI agents

Generate chat completions, speech, video, and music with the MiniMax models, and upload or list the files those generation jobs use.

## Scope

Does not train or fine-tune custom models, host a chat UI, or manage billing. Use it to run text, speech, video, and music generation jobs and to manage the files those jobs use.

## Capabilities

- Generate chat completions from the MiniMax large language models
- Synthesize speech from text with the text-to-audio endpoint
- Generate a video clip from a text prompt
- Generate a music track from a structured prompt
- Upload a file and list previously uploaded files

## Use cases

### Conversational Assistant

An agent answers a user's question by calling POST `/v1/text/chatcompletion_v2` with the conversation history, then returns the model's reply. This gives the assistant a language backend without wiring a separate provider.

Example prompt: Send the running conversation to POST `/v1/text/chatcompletion_v2` and return the assistant message

### Voiceover Generation

A content agent turns a written script into narration by calling POST `/v1/t2a_v2`, receiving audio it can attach to a video or podcast. It removes the manual step of recording or sourcing a voice.

Example prompt: Send the script text to POST `/v1/t2a_v2` and store the returned audio

### Short-Form Video Creation

A marketing agent produces a short clip from a scene description with POST `/v1/video_generation`, then polls for the result. This lets an automated pipeline draft video assets from a text brief.

Example prompt: Submit the scene description to POST `/v1/video_generation` and retrieve the generated clip

### Generation Asset Management

Before a batch of jobs, an agent uploads reference material with the file upload endpoint and confirms what is available by calling GET `/v1/files/list.` This keeps the inputs a generation run depends on organized.

Example prompt: Upload the reference file, then call GET `/v1/files/list` to confirm it is present

## Key endpoints

| Method | Path | Description |
| --- | --- | --- |
| POST | `/v1/text/chatcompletion_v2` | Generate a chat completion |
| POST | `/v1/t2a_v2` | Synthesize speech from text |
| POST | `/v1/video_generation` | Generate a video from a prompt |
| POST | `/v1/music_generation` | Generate music from a prompt |
| GET | `/v1/files/list` | List uploaded files |

## Key resources

- **Chat Completions** — Generate text replies from the MiniMax language models
- **Audio, Video, and Music** — Synthesize speech and generate video and music from prompts
- **Files** — Upload and list the files generation jobs use

## Why Jentic

- **Setup:** Wiring the MiniMax API by hand means setting up its bearer token, formatting a different request body for each of the text, speech, video, and music endpoints, and polling long-running jobs against https://api.minimaxi.com. Through Jentic you install once, import the MiniMax API from the API Directory, store the credential once, and your agent calls it.
- **Permission scoping:** Because you run Jentic One yourself, you decide which MiniMax operations the agent may call. Allow only the ones it needs, such as chat completion and file listing, and leave video and music generation out of the allowed set unless you add them.
- **Credential handling:** Your MiniMax bearer token is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context.
- **Discovery method:** Agents search Jentic by intent, such as 'generate a chat completion' or 'synthesize speech from text', and Jentic returns the matching MiniMax operation with its input schema so the agent calls the right endpoint.

## Related APIs

- **OpenAI API** — OpenAI offers text, image, and audio models; MiniMax adds video and music generation from one provider.
- **Stability AI API** — Stability AI specializes in image generation; MiniMax covers text, speech, video, and music.
- **ElevenLabs API** — ElevenLabs is a dedicated speech provider; MiniMax bundles speech with text, video, and music.

## FAQ

### What can an AI agent do with the MiniMax API?

An agent can generate a chat completion, synthesize speech from text, generate a video or a music track from a prompt, and upload or list the files those jobs use. The models are reached over one REST interface, so a single integration covers several media types.

### What authentication does the MiniMax API use?

The MiniMax API authenticates with a bearer token, supplied in the Authorization header as declared by the specification's bearerAuth scheme. Through Jentic the token is held by your own instance and attached to each request at execution time.

### Can I generate speech through the MiniMax API?

Yes. POST `/v1/t2a_v2` turns a block of text into audio you can attach to a video, podcast, or notification. Pair it with the chat completion endpoint to draft and voice a script in one workflow.

### Is there an MCP server for the MiniMax API?

You do not need a separate MCP server to use the MiniMax API with an agent. Through Jentic One, the self-hosted execution layer, your agent calls the MiniMax operations directly after you import them from the API Directory. Install Jentic One from github.com/jentic/jentic-one to run it yourself.

### How does my agent discover MiniMax operations through Jentic?

Agents search Jentic by intent, such as 'generate a chat completion with MiniMax', and Jentic returns the matching operation with its input schema. The agent then calls the right endpoint without browsing the reference docs.

### Can I limit what my agent is allowed to do with the MiniMax API?

Yes. Because you run Jentic One yourself, you decide which MiniMax operations the agent may call. You can allow only chat completion, for example, and leave video and music generation out of the allowed set unless you add them, so the agent acts only within the operations you permit.