MiniMax M3
MiniMax M3 is a large language model made by MiniMax, added to the catalog on 31 May 2026. It has a context window of 1,048,576 tokens and costs $0.30 per 1M input tokens and $1.20 per 1M output tokens. It accepts text, images and video and produces text. Typical uses: working with text and visual inputs, coding tasks and long-horizon agentic work.
Price checked ยท source: OpenRouter
What MiniMax M3 is
MiniMax M3 is a multimodal foundation model from MiniMax. It accepts text, image, and video inputs and produces text. The model is described as suited for coding and long-horizon agentic work. Its supported capabilities also include reasoning, tool use, structured outputs, and response formats. The listed input modalities do not include audio, and its listed output modality is text.
The context window is 1,048,576 tokens, while the maximum output is 512,000 tokens. This large context can hold extensive input material for a task, and the output limit allows for lengthy text responses. The provided facts do not specify separate context limits for text, images, or video, or explain how those modalities are represented within the token window. They also do not describe particular tools or response formats.
These features make MiniMax M3 relevant to tasks involving substantial context, coding, or long-horizon agentic work. It can take text alongside image or video inputs and return text, so it may fit workflows that involve examining more than written material. Tool support can be used in workflows that call tools, and structured-output and response-format support allow requests to specify how text should be returned. Its reasoning support is another listed capability for tasks that involve working through information. The model's documented modalities and features describe its general scope, rather than guaranteeing a particular result for any task.
Key features
- Accepts text, image, and video inputs
- Produces text output
- Has a 1,048,576-token context window
- Supports tools, reasoning, structured outputs, and response formats
Typical uses
- Working with text and visual inputs
- Coding tasks
- Long-horizon agentic work
- Generating structured text outputs
MiniMax M3 pricing
US dollars, list price through OpenRouter.
| Item | Standard | Batch |
|---|---|---|
| Input, per 1M tokens | $0.30 | $0.224 |
| Output, per 1M tokens | $1.20 | $0.88 |
| Cached input read, per 1M tokens | $0.06 | $0.0448 |
| Variant | Model id |
|---|---|
| Batch | minimax/minimax-m3:batch |
A variant is the same model under another id, with prices or limits of its own. Its prices are the columns of the list above.
No price change since it was added on 31 May 2026.
| What a task costs | Cost |
|---|---|
| Answer a short question500 in / 300 out | $0.0005 |
| Summarise a 50-page PDF25,000 in / 800 out | $0.0085 |
| Review a 2,000-line source file30,000 in / 2,000 out | $0.011 |
| Read a 300-page contract set150,000 in / 1,500 out | $0.047 |
MiniMax M3 specifications
| Model id | minimax/minimax-m3 |
|---|---|
| Version | minimax-m3-20260531 |
| Context window | 1,048,576 tokensAbout 786,432 English words |
| Max output | 512,000 tokens |
| Knowledge cutoff | Not published |
| Added to catalog | 31 May 2026 |
| Input types | text, image, video |
|---|---|
| Output types | text |
| Tool calling | Yes |
| Structured outputs | Yes |
| Reasoning effort control | No |
| Content moderation | Not applied by provider |
Sample outputs
The same prompts are given to every model, and the unedited output is shown with the date it was produced. Upend does not publish leaderboard scores it did not run.
No output has been recorded for MiniMax M3 yet.
Compare MiniMax M3
Side by side with other models: prices, context window, what each accepts and supports, and the same four tasks priced on both.