GPT Image 2.5 Flare & Sunburst are live on EvoLinkTry GPT Image 2.5

MiniMax M3.1 Flash · Prompt Hub

MiniMax M3.1 Flash Prompts & Use Cases

Browse 3D, game and website examples with their original prompts and matching previews.

Copy a complete prompt, prepare its reference assets, and use MiniMax M3.1 Flash through EvoLink’s unified API.

MiniMax M3.1 Flash Prompts & Use Cases — Prompt templates

Browse by workflow

MiniMax M3.1 Flash Prompt Library

14 / 1414 cases with original prompts, screenshots and videos for 3D, games and websites.
Temple of Heaven in Three.js
3D & Design

Temple of Heaven in Three.js

Model the three roof tiers, circular hall and terraces with interactive camera controls.

Millennium Falcon voxel model
3D & Design

Millennium Falcon voxel model

Rebuild the iconic spacecraft with grayscale voxels and an exploded view.

Experience Curiosity
3D & Design

Experience Curiosity

Explore Mars with a six-wheel rover, mast camera, robotic arm and mission replay.

Ancient palace structure
3D & Design

Ancient palace structure

Explore a Chinese palace structure through multiple interactive 3D viewing modes.

Y2K dance arcade
Games

Y2K dance arcade

Build a browser rhythm game with song selection, four-lane notes, pixel dancers and scoring.

Multiplayer flight chess
Games

Multiplayer flight chess

Implement local multiplayer and computer opponents with three board themes and complete rules.

Flappy Pet · Bamboo Hop
Games

Flappy Pet · Bamboo Hop

Create a flying-pet game with character selection, obstacles, collectibles and responsive controls.

2D city escape
Games

2D city escape

Escape a city while managing stamina, evading enemies and collecting three keys.

Restaurant menu and reservations
Web Design

Restaurant menu and reservations

Build a retro Y2K restaurant website with animated lettering, a menu and reservation pages.

Fragrance shopping website
Web Design

Fragrance shopping website

Compose a cinematic product website with chapter-based storytelling and independent video playback.

GTA character personality quiz
Web Design

GTA character personality quiz

Combine a first-person safehouse, twelve questions and shareable character results.

MiniMax H3 × M3.1 cinematic website
Web Design

MiniMax H3 × M3.1 cinematic website

Create a continuous cinematic website whose video timeline follows scrolling.

Monochrome creative portfolio
Web Design

Monochrome creative portfolio

Build a terminal-inspired creative archive with filters, video previews and project detail views.

Windows desktop simulation
Web Design

Windows desktop simulation

Recreate a browser desktop with window management, local apps, games and a music player.

Powered by EvoLink

How to write better MiniMax M3.1 Flash prompts

Describe the scene and controls: 3D proportions and camera views, game rules and win conditions, or website pages and navigation.

Provide reference photos, asset manifests, sprites, audio and video. List missing files before implementation.

Specify the stack and delivery: Three.js, Canvas or React; complete source files, dependencies, local startup and deployment steps. Use a coding environment that can write files and run commands.

Test the result in a browser: camera controls, collision and scoring, navigation and video playback, desktop and touch layouts. Report what was actually tested.

MiniMax M3.1 Flash Prompt FAQ

MiniMax M3.1 Flash Prompt FAQ

What is MiniMax M3.1 Flash good at?

It is designed for reasoning-heavy agent tasks, coding, long-context analysis, tool use, and multimodal inputs. These templates focus on workflows where a structured output and validation step matter.

Can reasoning be disabled?

M3.1 Flash uses deep thinking by default. Use a supported effort level such as low, medium, high, xhigh, or max instead of sending a disabled or none value.

Which reasoning field should an API request use?

Use the field for the API surface you call: reasoning_effort for OpenAI-compatible requests, output_config.effort for Anthropic-compatible requests, or reasoning.effort for Responses requests.

Can these prompts use images or video?

Yes. The model supports text, image, and video inputs. Keep file sizes and request limits in mind, and state which frame or image supports each visual conclusion.