Forge is available: build a website with AI from your RodiumAi account.

Try Forge
MiniMax

MiniMax M3

Healthychattoolsvision
Context
1M
Max output
-
Speed
Fast
Quality
Basic
Released
2026-05-31
Input: / Output:
ImageTextVideoText

About this model

Frontier multimodal agentic model with 1M context for coding, long-horizon tasks, and visual understanding. MiniMax M3 is MiniMax's new flagship open-weight model for coding, agentic execution, and multimodal reasoning.

Pricing

RateRODIUSD (ref.)Unit
Text tokens
In227.3~ 0.300per 1M tokens (RODI/M, USD/M)
Out909.1~ 1.20per 1M tokens (RODI/M, USD/M)
Prompt caching
Cached45.5~ 0.0600per 1M tokens (RODI/M, USD/M)

Pricing tiers

The rate depends on how you call the model. Each table below lists every tier, in RODI with the approximate USD value.

Long context pricing

When a prompt is longer than 512K tokens, the whole request is billed at these rates (RODI per 1M tokens).

Prompt over 512K tokens
  • In 454.6
  • Out 1,818
  • Cached 91

RODI prices include RodiumAi markup and upstream fees. USD figures are wholesale reference rates.

Providers & pricing

ProviderInputOutputCachedFree tierStatus
Runware

Runware

Default

All rates (3)
  • In227.3 RODI/M
  • Out909.1 RODI/M
  • Cached45.5 RODI/M
227.3909.145.5-No data

Capabilities

Streaming
Tool calling
Vision
JSON mode
Reasoning

API usage

Use the canonical slug with client.chat() or chat.completions.create().

Examples

…

Example response

…

RodiumAi SDK

Official Python, JavaScript, and Laravel SDKs: flat API aligned with this model slug.

…

Use cases

  • Read receipts, invoices, ID cards and forms, then return structured JSON.
  • Describe photos for accessibility or catalogue listings.
  • Check product photos against a checklist (damage, label, packaging).
  • Understand screenshots, charts and diagrams in support tickets.
  • Compare two images in a single request by sending several image_url parts.

Notes

  • Endpoint: POST https://api.rodiumai.io/v1/chat/completions, with content as a list of text and image_url parts.
  • image_url.url accepts an https:// URL or a base64 data URL (data:image/png;base64,...). A data URL with a missing or non-image MIME type is rejected with a 400 before reaching the model.
  • Images are billed as input tokens. Large images cost more, so resize photos to what the task needs.
  • Put the question in a text part of the same message as the image.
View all