# Mage — Unlimited AI Image, Video & Audio Generator

## 🎨 Platform Overview

**Mage** is an unlimited AI image, video, and audio generator trusted by millions of creators. Create characters, generate images, and produce high-quality videos and audio fast with state-of-the-art AI models.

**🧑‍💻 NEW: The Mage API and MCP server are live!** Developers can generate images, video, and audio from code through the REST API at https://api.mage.space/v1, and AI assistants (Claude, ChatGPT, Claude Code, Grok, Cursor) can connect through the MCP server at https://mcp.mage.space/mcp. See [Mage for Developers](https://www.mage.space/developers) and the full developer section below.

**📦 NEW: Official SDKs and ComfyUI nodes!** Install the TypeScript SDK (`npm install @mage-space/sdk`) or the Python SDK (`pip install mage-space`) to call every Mage model in a few lines, or add the Mage nodes to ComfyUI (`comfy node install comfyui-mage`) to run Mango, Cherry, and Seed Audio in your workflows. See [Mage SDKs](https://www.mage.space/developers/sdks) and [Mage for ComfyUI](https://www.mage.space/developers/comfyui).

**🧩 NEW: Mage skills!** Free, open-source Agent Skills that teach Claude, Codex, Cursor, and other agents complete jobs with Mage: product photoshoots, marketplace images, ad variations, UGC and faceless videos, YouTube thumbnails, and every Mage app, from face swaps and relighting to video edits and lip sync. Add them with `npx skills add mage-space/skills` and see [Mage skills](https://www.mage.space/skills) and the skills section below.

**🔊 NEW: AI Audio Generation is here!** Mage now generates audio, powered by the new **Seed Audio** architecture. And **Mango 3 is live now** — the latest generation of Mage's flagship exclusive image model. **Cherry 2 Pro is live now** — the newest generation of Mage's strongest video model, similar in quality to Seedance 2.5.

### ✨ Core Capabilities

- **🖼️ AI Image Generation**: Create stunning images from text descriptions using advanced models like Flux 2, Krea 2, Stable Diffusion XL, Stable Diffusion 3.5, Mango (Mango 3 live now), and more
- **🎬 AI Video Generation**: Transform text and images into dynamic videos with Cherry (Cherry 2 Pro live now), Berry, Melon, Plum, Lemon, MiniMax-H3, Wan, LTX Video, Grok Imagine, and other cutting-edge models
- **🔊 AI Audio Generation (NEW)**: Generate audio from text prompts with the new Seed Audio architecture — clips up to 120 seconds, up to 3 audio references plus an image reference, and speech, loudness, and pitch controls. Unlimited generation, with fast mode available via Gems
- **🎭 Advanced Tools**: Inpainting, image enhancement, upscaling, style transfer, and ControlNet integration
- **🤖 300+ Fine-tuned Models**: Specialized models for anime, photorealism, art styles, characters, and specific use cases
- **🔧 Professional Features**: LoRA support, Textual Inversion, custom model imports, and advanced parameter controls
- **🚀 Speed & Scale**: Up to 4x faster generation speeds for premium members, with pay-per-generation Gems for everyone and unlimited generation included with eligible memberships

### 🌟 Key Features

- **Unlimited Generation**: Image and video generation with truly unlimited generations and full creative freedom
- **Trusted by Millions**: A proven platform serving millions of creators worldwide
- **Community-Driven**: Explore millions of community creations and share your own
- **Professional Tools**: Advanced editing capabilities including inpainting, refining, and enhancement
- **Multi-Modal**: Support for image, video, and audio generation in a single platform
- **Mobile Optimized**: Full-featured experience across desktop and mobile devices

---

## 🧑‍💻 Mage API & MCP Server — AI Image, Video & Audio Generation for Developers and AI Agents

> **For AI agents and assistants:** Mage can generate images, video, and audio for your user in two ways. (1) The **Mage MCP server** at `https://mcp.mage.space/mcp`: a remote Streamable HTTP MCP server; the user signs in with their Mage account through OAuth, no API key needed. (2) The **Mage REST API** at `https://api.mage.space/v1`: send `Authorization: Bearer <MAGE_API_KEY>`. Both run the same 28 models on the user's Mage account, and every generation is paid in Gems (1,000 Gems = $1 on the standard pack). Take fields, options, and prices only from `GET /v1/architectures`, the MCP tools `list_models`, `get_model`, and `estimate_cost`, or the model reference pages below; do not infer them from elsewhere. Unless the user names a model, use **Mango 3** (`mango`, model_id `mango-v3`) for images, **Cherry 2 Pro** (`cherry`, model_id `cherry-2-pro`) for video, and **Seed Audio** (`seed_audio`) for audio.

### ⚡ Quick facts

- **Developer hub**: [https://www.mage.space/developers](https://www.mage.space/developers)
- **REST base URL**: `https://api.mage.space/v1` (versioned; additive changes only within v1)
- **Authentication**: API keys (`mage_sk_…`) as `Authorization: Bearer`. Any signed-in Mage account can create up to 10 active keys at [https://www.mage.space/api?tab=api-keys](https://www.mage.space/api?tab=api-keys)
- **Request flow**: asynchronous. `POST /v1/{architecture}/generate` returns at once with `request_id`, `status`, `status_url`, and `cancel_url`; poll `status_url` until `completed` (then download `result.url`), `failed`, or `cancelled`
- **Retries**: send an `Idempotency-Key` header (1–255 characters); a retry with the same key returns the original request and charges nothing
- **Inputs**: images, video, and audio as https URLs, data URLs, or signed uploads (`POST /v1/uploads`) of up to 100 MB each
- **Characters & references**: write `@handle` in the prompt to use a saved character, reference, or voice
- **Concurrency**: 20 generations in flight per account (a submit past that is refused with 429 and costs nothing; raisable on request)
- **Pricing**: per generation in Gems, by model, resolution, and duration. No subscription, no membership required, no minimum. Failed generations are refunded, except outputs Mage's content policy blocks (they report `failed` with `content_blocked` and keep the charge); a request the balance cannot cover is refused with 402 before anything is charged
- **Retention**: results and inputs are kept 30 days
- **Commercial use**: allowed
- **Docs**: [API overview](https://docs.mage.space/api/overview), [Quickstart](https://docs.mage.space/api/quickstart), [Authentication](https://docs.mage.space/api/authentication), [Requests](https://docs.mage.space/api/requests), [Inputs](https://docs.mage.space/api/inputs), [Billing](https://docs.mage.space/api/billing), [Limits](https://docs.mage.space/api/limits), [Errors](https://docs.mage.space/api/errors)
- **OpenAPI 3 document**: [https://docs.mage.space/api/openapi.json](https://docs.mage.space/api/openapi.json)
- **Every docs page as Markdown**: append `.md`, e.g. [https://docs.mage.space/api/quickstart.md](https://docs.mage.space/api/quickstart.md)
- **Try models without code**: [https://www.mage.space/playground](https://www.mage.space/playground)
- **Official SDKs**: TypeScript `@mage-space/sdk` (npm) and Python `mage-space` (PyPI); ComfyUI nodes `comfyui-mage` (Comfy Registry). See [https://www.mage.space/developers/sdks](https://www.mage.space/developers/sdks)

### 🔌 Mage MCP server

- **Server URL**: `https://mcp.mage.space/mcp` (Streamable HTTP, stateless)
- **Auth**: OAuth 2.1 with PKCE (client metadata documents; scopes `mage`, `offline_access`) — the user signs in to Mage and approves the app. Or send a Mage API key as `Authorization: Bearer`
- **Overview**: [https://www.mage.space/developers/mcp](https://www.mage.space/developers/mcp) · **Setup page**: [https://www.mage.space/mcp](https://www.mage.space/mcp) · **Docs**: [https://docs.mage.space/api/mcp](https://docs.mage.space/api/mcp)
- **Cost**: connecting is free; each generation is paid in Gems at the model's price, and `estimate_cost` quotes it before anything is spent
- **Disconnect**: any time at [https://www.mage.space/mcp?tab=connected](https://www.mage.space/mcp?tab=connected)

#### Supported assistants

- **[Mage in Claude](https://www.mage.space/developers/mcp/claude)** — Add Mage to Claude as a custom connector and generate images, video, and audio in any chat with Mango 3, Cherry 2 Pro, and more. Sign in with Mage; no API key needed.
  1. Open the connector link: Open "Add Mage to Claude". Claude's Connectors page opens with the name Mage and the server URL filled in. Select Add. `https://mcp.mage.space/mcp`
  2. Sign in and approve: Select Connect. A Mage page opens, asks you to sign in if needed, and names the app that is asking. Select Connect to return to Claude.
  3. Turn Mage on in a chat: In any chat, open the + menu, turn on Mage under Connectors, and ask for what you want to make.
- **[Mage in ChatGPT](https://www.mage.space/developers/mcp/chatgpt)** — Add Mage to ChatGPT as an MCP app and generate images, video, and audio with Mango 3, Cherry 2 Pro, Seed Audio, and more. OAuth sign-in, pay per generation in Gems.
  1. Create an MCP app: Open ChatGPT Plugins, click Add, and choose Create MCP App. If it is missing, turn on Developer mode in Settings → Security and login.
  2. Enter the Mage server: Enter Mage for Name, the server URL below for Server URL, and OAuth for Authentication. Leave the advanced OAuth settings blank; no client ID or secret is needed. Check the notice and click Create. `https://mcp.mage.space/mcp`
  3. Sign in to Mage: Sign in on the Mage page that opens, check the app name, and select Connect.
  4. Use it in a chat: Install Mage from your Personal plugins, start a Work chat, and mention @Mage with what you want to create.
- **[Mage in Claude Code](https://www.mage.space/developers/mcp/claude-code)** — Add the Mage MCP server to Claude Code with one command and generate images, video, and audio for your project. Upload local files, sign in with OAuth or an API key.
  1. Add the server: Run this command in your terminal. `claude mcp add --transport http mage https://mcp.mage.space/mcp`
  2. Authenticate: In Claude Code, run /mcp, choose mage, and authenticate. A browser page opens for you to sign in to Mage and select Connect.
  3. Or use an API key: To skip the browser sign-in, add the server with your key in a header instead. `claude mcp add --transport http mage https://mcp.mage.space/mcp --header "Authorization: Bearer $MAGE_API_KEY"`
- **[Mage in Grok](https://www.mage.space/developers/mcp/grok)** — Connect Mage to Grok as a custom MCP connector and generate images, video, and audio with Mango 3, Cherry 2 Pro, and more, paid per generation in Gems.
  1. Create a custom connector: Open Grok connectors, select New Connector, and choose Custom.
  2. Enter the Mage server: Name it Mage, paste the server URL below, and follow the prompts to sign in to Mage and approve the connection. `https://mcp.mage.space/mcp`
  3. Ask Grok to use Mage: Start a conversation and ask Grok to use Mage to create an image, video, or audio.
- **[Mage in Cursor](https://www.mage.space/developers/mcp/cursor)** — Install the Mage MCP server in Cursor with one click and let the agent generate images, video, and audio for your app. OAuth with a public client ID, or an API key.
  1. Install with one click: Select "Add Mage to Cursor" to install the server with its public OAuth client ID. If Mage is already installed, replace its configuration.
  2. Or add it to mcp.json: For manual setup, add this entry to ~/.cursor/mcp.json. The client ID is public and needs no secret. `{ "mcpServers": { "mage": { "url": "https://mcp.mage.space/mcp", "auth": { "CLIENT_ID": "https://cursor.com/mcp", "scopes": [ "mage", "offline_access" ] } } } }`
  3. Sign in and ask: Authenticate with your Mage account when prompted, then ask Cursor in Agent mode to generate media for your project.

#### MCP tools (16)

- `list_models` — Lists the recommended models with prices, what each is best for, and its trade-offs.
- `get_model` — Describes one model's fields, allowed values, defaults, rules, and price.
- `estimate_cost` — Prices a generation before it starts. Nothing is charged.
- `create_upload` — Gets a signed link to upload a local file, for assistants that can run commands.
- `generate` — Starts an image, video, or audio generation.
- `get_request` — Waits for a generation to finish and returns the result link.
- `cancel_request` — Stops a running generation.
- `get_account` — Shows the connected account and its Gems balance.
- `list_characters` — Lists your saved characters and their @handles.
- `create_character` — Saves a private character from an image, optionally with a voice.
- `delete_character` — Deletes one of your characters.
- `list_references` — Lists your saved outfits, poses, locations, objects, and voice clips.
- `create_reference` — Saves a private reference from an image or audio link.
- `delete_reference` — Deletes one of your references.
- `list_history` — Lists your generations from the last 30 days, including ones made in the app.
- `search_creations` — Searches your saved creations by what they show, collection, or date.

### 📦 SDKs and ComfyUI

Prefer the official SDKs over hand-written HTTP: they submit, poll with backoff, send an `Idempotency-Key` on every submission (reused on their own retries, so a retry never charges twice), and upload files. All are MIT-licensed and in beta (0.x).

- **[TypeScript SDK](https://www.mage.space/developers/sdks/typescript)** — `npm install @mage-space/sdk` · [npm](https://www.npmjs.com/package/@mage-space/sdk) · [source](https://github.com/mage-space/mage-typescript) · [guide](https://docs.mage.space/api/sdks/typescript)
- **[Python SDK](https://www.mage.space/developers/sdks/python)** — `pip install mage-space` · [PyPI](https://pypi.org/project/mage-space/) · [source](https://github.com/mage-space/mage-python) · [guide](https://docs.mage.space/api/sdks/python)
- **[Mage for ComfyUI](https://www.mage.space/developers/comfyui)** — `comfy node install comfyui-mage` or search "Mage" in ComfyUI-Manager · [Comfy Registry](https://registry.comfy.org/nodes/comfyui-mage) · [source](https://github.com/mage-space/ComfyUI-Mage) · [guide](https://docs.mage.space/api/integrations/comfyui). Set `MAGE_API_KEY` before starting ComfyUI. Nodes: Mage Mango Image → IMAGE; Mage Cherry Video → VIDEO; Mage Seed Audio → AUDIO; Mage Generate (any model) → IMAGE, VIDEO, or AUDIO

```ts
import Mage from '@mage-space/sdk';

const mage = new Mage(); // reads MAGE_API_KEY

const request = await mage.run('mango', {
  prompt: 'Editorial portrait in soft daylight, 35mm film look',
  aspect_ratio: '4:5',
  model_id: 'mango-v3',
});

console.log(request.result.url);
```

```python
from mage_space import Mage

mage = Mage()  # reads MAGE_API_KEY

request = mage.run("mango", {
    "prompt": "Editorial portrait in soft daylight, 35mm film look",
    "aspect_ratio": "4:5",
    "model_id": "mango-v3",
})

print(request["result"]["url"])
```

### 🛠️ REST endpoints

- `POST /v1/{architecture}/generate` — Start a generation; returns request_id, status, status_url, cancel_url, billing.
- `GET /v1/requests/{request_id}/status` — Read a request: queued, in_progress, completed, failed, or cancelled; result.url when completed.
- `POST /v1/requests/{request_id}/cancel` — Cancel a running request (Gems are not refunded).
- `GET /v1/architectures` — The live model catalog: every architecture, its model_id values, options, defaults, input fields, and JSON Schema.
- `GET /v1/account` — The account and its Gems balance.
- `POST /v1/uploads` — A signed URL to upload an image, video, or audio input of up to 100 MB.
- `GET /v1/characters` — List saved characters and their @handles.
- `POST /v1/characters` — Save a character from an image, optionally with a voice clip.
- `DELETE /v1/characters/{character_id}` — Delete a character.
- `GET /v1/references` — List saved references (outfits, poses, locations, objects, voice clips).
- `POST /v1/references` — Save a reference from an image or audio URL.
- `DELETE /v1/references/{reference_id}` — Delete a reference.

### ▶️ Minimal example

```bash
curl --request POST 'https://api.mage.space/v1/mango/generate' \
  --header "Authorization: Bearer $MAGE_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
  "prompt": "Editorial portrait in soft daylight, 35mm film look",
  "aspect_ratio": "4:5",
  "model_id": "mango-v3"
}'
```

### Image generation models in the API (15)

- **[Anima API](https://www.mage.space/developers/models/anima)** — Purpose-built for anime. An anime and illustration model with clean lines and flat color, with Danbooru-style tags and prompt weighting.
  - Endpoint: `POST https://api.mage.space/v1/anima/generate`
  - model_id: `anima-v1` (Anima v1) default
  - Price at default settings: 20 Gems (~$0.02)
  - Media inputs: none (prompt only)
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/anima.md](https://docs.mage.space/api/models/anima.md)
- **[Chroma API](https://www.mage.space/developers/models/chroma)** — Open-weights, built on Flux. An open-weights general-purpose image model on the Flux architecture.
  - Endpoint: `POST https://api.mage.space/v1/chroma/generate`
  - model_id: `chroma-v1-hd` default
  - Price at default settings: 35 Gems (~$0.035)
  - Media inputs: none (prompt only)
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/chroma.md](https://docs.mage.space/api/models/chroma.md)
- **[Flux 2 API](https://www.mage.space/developers/models/flux-2)** — Long prompts and multi-reference edits. Flux 2 Dev, an open-weights model for long, detailed prompts, generation, and edits from up to three reference images.
  - Endpoint: `POST https://api.mage.space/v1/flux2/generate`
  - model_id: `flux2-dev` default
  - Price at default settings: 40 Gems (~$0.04)
  - Media inputs: `image`, `additional_images`
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/flux2.md](https://docs.mage.space/api/models/flux2.md)
- **[GPT Image 2 API](https://www.mage.space/developers/models/gpt-image-2)** — Precise prompts and readable text. OpenAI's GPT Image 2, GPT Image 2.5 Flare, and GPT Image 2.5 Sunburst, for generation and edits. OpenAI's safety filter applies.
  - Endpoint: `POST https://api.mage.space/v1/gpt_image_2/generate`
  - model_id: `gpt-image-2` (GPT Image 2) default, `gpt-image-2.5-flare` (GPT Image 2.5 Flare), `gpt-image-2.5-sunburst` (GPT Image 2.5 Sunburst)
  - Price at default settings: 9 Gems (~$0.009)
  - Media inputs: `image`, `additional_images`
  - @handle mentions: characters, references
  - Reference: [https://docs.mage.space/api/models/gpt_image_2.md](https://docs.mage.space/api/models/gpt_image_2.md)
- **[Grok Image API](https://www.mage.space/developers/models/grok-image)** — Grok Imagine, in three quality tiers. xAI's Grok Imagine image models in three quality tiers, for generation and edits from reference images. xAI's safety filter applies.
  - Endpoint: `POST https://api.mage.space/v1/grok_image/generate`
  - model_id: `grok-imagine-image` (Grok Image), `grok-imagine-image-quality` (Grok Image Quality) default, `grok-imagine-image-2.0` (Grok Image 2.0)
  - Price at default settings: 75 Gems (~$0.075)
  - Media inputs: `image`, `additional_images`
  - @handle mentions: characters, references
  - Reference: [https://docs.mage.space/api/models/grok_image.md](https://docs.mage.space/api/models/grok_image.md)
- **[Guava API](https://www.mage.space/developers/models/guava)** — Photorealism that passes for a photograph. Mage's photorealism family, for portraits, fashion, and product shots that pass for photographs, with reference images. Guava 2 and Guava 2 Pro are the current pair.
  - Endpoint: `POST https://api.mage.space/v1/guava/generate`
  - model_id: `guava` (Guava) default, `guava-pro` (Guava Pro), `guava-pro-v1-5` (Guava Pro 1.5), `guava-2` (Guava 2), `guava-2-pro` (Guava 2 Pro)
  - Price at default settings: 37 Gems (~$0.037)
  - Media inputs: `image`, `additional_images`
  - @handle mentions: characters, references
  - Reference: [https://docs.mage.space/api/models/guava.md](https://docs.mage.space/api/models/guava.md)
- **[HiDream API](https://www.mage.space/developers/models/hidream)** — Fast and versatile, with negative prompts. A fast general image model with a negative prompt, photoreal or illustrated.
  - Endpoint: `POST https://api.mage.space/v1/hidream/generate`
  - model_id: `hidream-fast` default
  - Price at default settings: 20 Gems (~$0.02)
  - Media inputs: none (prompt only)
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/hidream.md](https://docs.mage.space/api/models/hidream.md)
- **[Krea 2 API](https://www.mage.space/developers/models/krea-2)** — Gorgeous aesthetics in seconds. A fast image model with a strong built-in aesthetic, at 1K or 2K.
  - Endpoint: `POST https://api.mage.space/v1/krea_2/generate`
  - model_id: `krea-2-turbo` (Krea 2 Turbo) default
  - Price at default settings: 20 Gems (~$0.02)
  - Media inputs: none (prompt only)
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/krea_2.md](https://docs.mage.space/api/models/krea_2.md)
- **[Mango API](https://www.mage.space/developers/models/mango)** — Our flagship image family. Mage's flagship image family, for consistent characters, precise edits, and scenes merged from many references. Mango 3 (`mango-v3`) is the current model; Mango 3 Turbo (`mango-v3-turbo`) is the fastest and cheapest, and Mango 3S (`mango-v3s`) reaches 3K.
  - Endpoint: `POST https://api.mage.space/v1/mango/generate`
  - model_id: `mango` (Mango), `mango-v2` (Mango 2), `mango-v3s` (Mango 3S), `mango-v3` (Mango 3) default, `mango-v3-turbo` (Mango 3 Turbo)
  - Price at default settings: 135 Gems (~$0.135)
  - Media inputs: `image`, `additional_images`
  - @handle mentions: characters, references
  - Reference: [https://docs.mage.space/api/models/mango.md](https://docs.mage.space/api/models/mango.md)
- **[Nano Banana 2 API](https://www.mage.space/developers/models/nano-banana-2)** — Crisp, readable text, up to 4K. Google's Nano Banana models, for graphics with readable text and edits from up to 14 reference images, up to 4K. Nano Banana 2.1 (`nano-banana-v2.1`), the default, costs less; Nano Banana 2 (`nano-banana-v2`) also renders at 512. Google's safety filter applies.
  - Endpoint: `POST https://api.mage.space/v1/nano_banana_v2/generate`
  - model_id: `nano-banana-v2` (Nano Banana 2), `nano-banana-v2.1` (Nano Banana 2.1) default
  - Price at default settings: 51 Gems (~$0.051)
  - Media inputs: `image`, `additional_images`
  - @handle mentions: characters, references
  - Reference: [https://docs.mage.space/api/models/nano_banana_v2.md](https://docs.mage.space/api/models/nano_banana_v2.md)
- **[SDXL Plus API](https://www.mage.space/developers/models/sdxl-plus)** — SDXL with faces and hands fixed. Mage's high-detail SDXL pipeline with automatic face and hand cleanup, with an optional input image.
  - Endpoint: `POST https://api.mage.space/v1/sdxl_plus/generate`
  - model_id: `6A35A7855770AE9820A3C931D4964C3817B6D9E3C6F9C4DABB5B3A94E5643B80` default
  - Price at default settings: 15 Gems (~$0.015)
  - Media inputs: `image`
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/sdxl_plus.md](https://docs.mage.space/api/models/sdxl_plus.md)
- **[Stable Diffusion v1.5 API](https://www.mage.space/developers/models/stable-diffusion-v1-5)** — The classic, on a budget. The classic Stable Diffusion 1.5, with an optional input image.
  - Endpoint: `POST https://api.mage.space/v1/stable_diffusion_v15/generate`
  - model_id: `15012C538F503CE2EBFC2C8547B268C75CCDAFF7A281DB55399940FF1D70E21D` default
  - Price at default settings: 10 Gems (~$0.01)
  - Media inputs: `image`
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/stable_diffusion_v15.md](https://docs.mage.space/api/models/stable_diffusion_v15.md)
- **[Stable Diffusion v3.5 Large API](https://www.mage.space/developers/models/stable-diffusion-v3-5-large)** — The Stable Diffusion that listens best. Stable Diffusion 3.5 Large, the Stable Diffusion that follows detailed prompts most closely, with an optional input image.
  - Endpoint: `POST https://api.mage.space/v1/stable_diffusion_v35_large/generate`
  - model_id: `sd-3-5-large` default
  - Price at default settings: 10 Gems (~$0.01)
  - Media inputs: `image`
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/stable_diffusion_v35_large.md](https://docs.mage.space/api/models/stable_diffusion_v35_large.md)
- **[Stable Diffusion XL API](https://www.mage.space/developers/models/stable-diffusion-xl)** — Photoreal-tuned SDXL. Stable Diffusion XL, tuned for photoreal output, with an optional input image.
  - Endpoint: `POST https://api.mage.space/v1/stable_diffusion_xl/generate`
  - model_id: `6A35A7855770AE9820A3C931D4964C3817B6D9E3C6F9C4DABB5B3A94E5643B80` default
  - Price at default settings: 10 Gems (~$0.01)
  - Media inputs: `image`
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/stable_diffusion_xl.md](https://docs.mage.space/api/models/stable_diffusion_xl.md)
- **[Z-Image API](https://www.mage.space/developers/models/z-image)** — The fastest image model on Mage. The fastest image model on Mage: photoreal text-to-image with solid anatomy and lettering, at 1K or 2K.
  - Endpoint: `POST https://api.mage.space/v1/z_image/generate`
  - model_id: `z-image-turbo` (Z-Image) default
  - Price at default settings: 10 Gems (~$0.01)
  - Media inputs: none (prompt only)
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/z_image.md](https://docs.mage.space/api/models/z_image.md)

### Video generation models in the API (12)

- **[Berry API](https://www.mage.space/developers/models/berry)** — Video that stays true to your references. Mage's reference-driven video model, faithful to the images you give it; Berry 2 is the current version. A first frame or reference images, or a source video to edit.
  - Endpoint: `POST https://api.mage.space/v1/berry/generate`
  - model_id: `berry` (Berry), `berry-2` (Berry 2) default
  - Price at default settings: 189 Gems (~$0.189)
  - Media inputs: `first_image`, `image`, `additional_images`, `video`
  - @handle mentions: characters, references
  - Reference: [https://docs.mage.space/api/models/berry.md](https://docs.mage.space/api/models/berry.md)
- **[Blueberry API](https://www.mage.space/developers/models/blueberry)** — Pin the first and last frame. A video model that fills the motion between a first and a last frame, with audio; Blueberry 2 is the current version.
  - Endpoint: `POST https://api.mage.space/v1/blueberry/generate`
  - model_id: `blueberry`, `blueberry-v2` default
  - Price at default settings: 360 Gems (~$0.36)
  - Media inputs: `first_image`, `last_image`
  - @handle mentions: audio references
  - Reference: [https://docs.mage.space/api/models/blueberry.md](https://docs.mage.space/api/models/blueberry.md)
- **[Cherry API](https://www.mage.space/developers/models/cherry)** — Our flagship video family. Mage's flagship video line, with cinematic motion and audio: Cherry Mini, Cherry, Cherry Pro, and Cherry 2 Pro, the most powerful, with clips up to 30 seconds.
  - Endpoint: `POST https://api.mage.space/v1/cherry/generate`
  - model_id: `cherry-mini` (Cherry Mini), `cherry` (Cherry), `cherry-pro` (Cherry Pro), `cherry-2-pro` (Cherry 2 Pro) default
  - Price at default settings: 618 Gems (~$0.618)
  - Media inputs: `image`, `additional_images`, `videos`
  - @handle mentions: characters, references, audio references
  - Reference: [https://docs.mage.space/api/models/cherry.md](https://docs.mage.space/api/models/cherry.md)
- **[Grok Video API](https://www.mage.space/developers/models/grok-video)** — Grok Imagine, in motion. xAI's Grok Imagine for video, from text, reference images, or a source video to edit. xAI's safety filter applies.
  - Endpoint: `POST https://api.mage.space/v1/grok_video/generate`
  - model_id: `grok-imagine-video` (Grok Video) default
  - Price at default settings: 375 Gems (~$0.375)
  - Media inputs: `image`, `additional_images`, `video`
  - @handle mentions: characters, references
  - Reference: [https://docs.mage.space/api/models/grok_video.md](https://docs.mage.space/api/models/grok_video.md)
- **[Kiwi API](https://www.mage.space/developers/models/kiwi)** — Everyday video, zero setup. Everyday video from text or a single start frame, with nothing else to set up.
  - Endpoint: `POST https://api.mage.space/v1/kiwi/generate`
  - model_id: `kiwi` default
  - Price at default settings: 270 Gems (~$0.27)
  - Media inputs: `first_image`
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/kiwi.md](https://docs.mage.space/api/models/kiwi.md)
- **[Lemon API](https://www.mage.space/developers/models/lemon)** — Every video feature, at a friendlier price. Mage's all-round video model at a friendlier price: frames, references, Voices, generated audio, and clips from 2 to 30 seconds.
  - Endpoint: `POST https://api.mage.space/v1/lemon/generate`
  - model_id: `lemon` default
  - Price at default settings: 245 Gems (~$0.245)
  - Media inputs: `first_image`, `last_image`, `image`, `additional_images`, `videos`
  - @handle mentions: characters, references, audio references
  - Reference: [https://docs.mage.space/api/models/lemon.md](https://docs.mage.space/api/models/lemon.md)
- **[LTX Video API](https://www.mage.space/developers/models/ltx-video)** — Fast open-weights video for iteration. A fast open-weights video model for cheap iteration, in Distilled and Dev variants, from text or a start frame.
  - Endpoint: `POST https://api.mage.space/v1/ltx_video/generate`
  - model_id: `ltx-video-096-distilled` default, `ltx-video-096-dev`
  - Price at default settings: 20 Gems (~$0.02)
  - Media inputs: `first_image`
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/ltx_video.md](https://docs.mage.space/api/models/ltx_video.md)
- **[Melon API](https://www.mage.space/developers/models/melon)** — Anime in motion, with optional audio. Mage's anime video model, as Melon and Melon Pro, with optional audio.
  - Endpoint: `POST https://api.mage.space/v1/melon/generate`
  - model_id: `melon` (Melon) default, `melon-pro` (Melon Pro)
  - Price at default settings: 147 Gems (~$0.147)
  - Media inputs: `first_image`, `last_image`, `image`, `additional_images`
  - @handle mentions: characters, references
  - Reference: [https://docs.mage.space/api/models/melon.md](https://docs.mage.space/api/models/melon.md)
- **[MiniMax-H3 API](https://www.mage.space/developers/models/minimax-h3)** — Fast open video with stereo sound. A fast open video model that generates video and stereo audio together.
  - Endpoint: `POST https://api.mage.space/v1/minimax_h3/generate`
  - model_id: `minimax-h3-turbo` (MiniMax-H3 Turbo) default, `minimax-h3` (MiniMax-H3)
  - Price at default settings: 30 Gems (~$0.03)
  - Media inputs: `first_image`, `last_image`, `image`, `additional_images`, `videos`
  - @handle mentions: characters, references, audio references
  - Reference: [https://docs.mage.space/api/models/minimax_h3.md](https://docs.mage.space/api/models/minimax_h3.md)
- **[Plum API](https://www.mage.space/developers/models/plum)** — Sound on every clip, up to 2K. Mage's full-feature video model with generated audio on every clip and output up to 2K, as Plum and the faster Plum Max.
  - Endpoint: `POST https://api.mage.space/v1/plum/generate`
  - model_id: `plum` (Plum), `plum-max` (Plum Max) default
  - Price at default settings: 338 Gems (~$0.338)
  - Media inputs: `first_image`, `last_image`, `image`, `additional_images`, `videos`
  - @handle mentions: characters, references, audio references
  - Reference: [https://docs.mage.space/api/models/plum.md](https://docs.mage.space/api/models/plum.md)
- **[Raspberry API](https://www.mage.space/developers/models/raspberry)** — Characters brought to life from a first frame. A character video model from a first frame and reference images, with Voices and audio.
  - Endpoint: `POST https://api.mage.space/v1/raspberry/generate`
  - model_id: `raspberry` default
  - Price at default settings: 360 Gems (~$0.36)
  - Media inputs: `first_image`, `image`, `additional_images`
  - @handle mentions: characters, references
  - Reference: [https://docs.mage.space/api/models/raspberry.md](https://docs.mage.space/api/models/raspberry.md)
- **[Wan Video v2.2 API](https://www.mage.space/developers/models/wan-video-v2-2)** — Open-weights video with a Lightning variant. Wan Video v2.2, an open-weights video model from text or a start frame, with a Lightning variant.
  - Endpoint: `POST https://api.mage.space/v1/wan_22/generate`
  - model_id: `wan22-video` default, `wan22-video-lightning`
  - Price at default settings: 40 Gems (~$0.04)
  - Media inputs: `first_image`
  - @handle mentions: none
  - Reference: [https://docs.mage.space/api/models/wan_22.md](https://docs.mage.space/api/models/wan_22.md)

### Audio generation models in the API (1)

- **[Seed Audio API](https://www.mage.space/developers/models/seed-audio)** — Voices, music, and sound effects from one prompt. Voices, dialogue, music, and sound effects from one prompt, as an MP3 of exactly the requested duration.
  - Endpoint: `POST https://api.mage.space/v1/seed_audio/generate`
  - model_id: `seed-audio-1.0` (Seed Audio) default
  - Price at default settings: 19 Gems (~$0.019)
  - Media inputs: `image`
  - @handle mentions: audio references
  - Reference: [https://docs.mage.space/api/models/seed_audio.md](https://docs.mage.space/api/models/seed_audio.md)

### 📄 Developer pages

- **[Developer API & MCP Server for AI Image, Video & Audio](https://www.mage.space/developers)** — Build with Mage: one REST API and one MCP server for 28 AI image, video, and audio models, including Mango 3 and Cherry 2 Pro. Pay per generation in Gems.
- **[AI Image Generation API — Mango 3, Flux 2, Nano Banana 2](https://www.mage.space/developers/ai-image-generation-api)** — One REST API for 15 AI image models: text-to-image, image editing, consistent characters, and up to 4K output. Pay per image in Gems, no subscription.
- **[AI Video Generation API — Text-to-Video & Image-to-Video](https://www.mage.space/developers/ai-video-generation-api)** — Generate AI video from text, images, first and last frames, or references with 12 models including Cherry 2 Pro. One REST API, pay per clip in Gems.
- **[AI Audio Generation API — Voices, Music & Sound Effects](https://www.mage.space/developers/ai-audio-generation-api)** — Generate speech, voices, music, and sound effects from a text prompt with Seed Audio. Clips up to 120 seconds, voice references, one REST API, pay per clip in Gems.
- **[AI Model APIs — 28 Image, Video & Audio Models](https://www.mage.space/developers/models)** — Every model in the Mage API with its model_id values and price: Mango 3, Cherry 2 Pro, Nano Banana 2, GPT Image 2.5, Flux 2, Seed Audio, and 22 more.
- **[Anima API: Anima v1 Image Generation](https://www.mage.space/developers/models/anima)** — Anima API: purpose-built for anime. From 20 Gems ($0.02) per image, no subscription, via REST or MCP.
- **[Chroma API: Image Generation](https://www.mage.space/developers/models/chroma)** — Chroma API: open-weights, built on Flux. From 35 Gems ($0.035) per image, no subscription, via REST or MCP.
- **[Flux 2 API: Image Generation](https://www.mage.space/developers/models/flux-2)** — Flux 2 API: long prompts and multi-reference edits. From 40 Gems ($0.04) per image, no subscription, via REST or MCP.
- **[GPT Image 2 API: Image Generation](https://www.mage.space/developers/models/gpt-image-2)** — GPT Image 2 API: precise prompts and readable text. Models: GPT Image 2, GPT Image 2.5 Flare, GPT Image 2.5 Sunburst. From 9 Gems ($0.009) per image, no subscription, via REST or MCP.
- **[Grok Image API: Grok Image Quality Image Generation](https://www.mage.space/developers/models/grok-image)** — Grok Image API: grok Imagine, in three quality tiers. Models: Grok Image, Grok Image Quality, Grok Image 2.0. From 75 Gems ($0.075) per image, no subscription, via REST or MCP.
- **[Guava API: Image Generation](https://www.mage.space/developers/models/guava)** — Guava API: photorealism that passes for a photograph. Models: Guava, Guava Pro, Guava Pro 1.5, Guava 2, Guava 2 Pro. From 37 Gems ($0.037) per image, no subscription, via REST or MCP.
- **[HiDream API: Image Generation](https://www.mage.space/developers/models/hidream)** — HiDream API: fast and versatile, with negative prompts. From 20 Gems ($0.02) per image, no subscription, via REST or MCP.
- **[Krea 2 API: Krea 2 Turbo Image Generation](https://www.mage.space/developers/models/krea-2)** — Krea 2 API: gorgeous aesthetics in seconds. From 20 Gems ($0.02) per image, no subscription, via REST or MCP.
- **[Mango API: Mango 3 Image Generation](https://www.mage.space/developers/models/mango)** — Mango API: our flagship image family. Models: Mango, Mango 2, Mango 3S, Mango 3, Mango 3 Turbo. From 135 Gems ($0.135) per image, no subscription, via REST or MCP.
- **[Nano Banana 2 API: Nano Banana 2.1 Image Generation](https://www.mage.space/developers/models/nano-banana-2)** — Nano Banana 2 API: crisp, readable text, up to 4K. Models: Nano Banana 2, Nano Banana 2.1. From 51 Gems ($0.051) per image, no subscription, via REST or MCP.
- **[SDXL Plus API: Image Generation](https://www.mage.space/developers/models/sdxl-plus)** — SDXL Plus API: sDXL with faces and hands fixed. From 15 Gems ($0.015) per image, no subscription, via REST or MCP.
- **[Stable Diffusion v1.5 API: Image Generation](https://www.mage.space/developers/models/stable-diffusion-v1-5)** — Stable Diffusion v1.5 API: the classic, on a budget. From 10 Gems ($0.01) per image, no subscription, via REST or MCP.
- **[Stable Diffusion v3.5 Large API: Image Generation](https://www.mage.space/developers/models/stable-diffusion-v3-5-large)** — Stable Diffusion v3.5 Large API: the Stable Diffusion that listens best. From 10 Gems ($0.01) per image, no subscription, via REST or MCP.
- **[Stable Diffusion XL API: Image Generation](https://www.mage.space/developers/models/stable-diffusion-xl)** — Stable Diffusion XL API: photoreal-tuned SDXL. From 10 Gems ($0.01) per image, no subscription, via REST or MCP.
- **[Z-Image API: Image Generation](https://www.mage.space/developers/models/z-image)** — Z-Image API: the fastest image model on Mage. From 10 Gems ($0.01) per image, no subscription, via REST or MCP.
- **[Berry API: Berry 2 Video Generation](https://www.mage.space/developers/models/berry)** — Berry API: video that stays true to your references. Models: Berry, Berry 2. From 189 Gems ($0.189) per clip, no subscription, via REST or MCP.
- **[Blueberry API: Video Generation](https://www.mage.space/developers/models/blueberry)** — Blueberry API: pin the first and last frame. From 360 Gems ($0.36) per clip, no subscription, via REST or MCP.
- **[Cherry API: Cherry 2 Pro Video Generation](https://www.mage.space/developers/models/cherry)** — Cherry API: our flagship video family. Models: Cherry Mini, Cherry, Cherry Pro, Cherry 2 Pro. From 618 Gems ($0.618) per clip, no subscription, via REST or MCP.
- **[Grok Video API: Video Generation](https://www.mage.space/developers/models/grok-video)** — Grok Video API: grok Imagine, in motion. From 375 Gems ($0.375) per clip, no subscription, via REST or MCP.
- **[Kiwi API: Video Generation](https://www.mage.space/developers/models/kiwi)** — Kiwi API: everyday video, zero setup. From 270 Gems ($0.27) per clip, no subscription, via REST or MCP.
- **[Lemon API: Video Generation](https://www.mage.space/developers/models/lemon)** — Lemon API: every video feature, at a friendlier price. From 245 Gems ($0.245) per clip, no subscription, via REST or MCP.
- **[LTX Video API: Video Generation](https://www.mage.space/developers/models/ltx-video)** — LTX Video API: fast open-weights video for iteration. From 20 Gems ($0.02) per clip, no subscription, via REST or MCP.
- **[Melon API: Video Generation](https://www.mage.space/developers/models/melon)** — Melon API: anime in motion, with optional audio. Models: Melon, Melon Pro. From 147 Gems ($0.147) per clip, no subscription, via REST or MCP.
- **[MiniMax-H3 API: MiniMax-H3 Turbo Video Generation](https://www.mage.space/developers/models/minimax-h3)** — MiniMax-H3 API: fast open video with stereo sound. Models: MiniMax-H3 Turbo, MiniMax-H3. From 30 Gems ($0.03) per clip, no subscription, via REST or MCP.
- **[Plum API: Plum Max Video Generation](https://www.mage.space/developers/models/plum)** — Plum API: sound on every clip, up to 2K. Models: Plum, Plum Max. From 338 Gems ($0.338) per clip, no subscription, via REST or MCP.
- **[Raspberry API: Video Generation](https://www.mage.space/developers/models/raspberry)** — Raspberry API: characters brought to life from a first frame. From 360 Gems ($0.36) per clip, no subscription, via REST or MCP.
- **[Wan Video v2.2 API: Video Generation](https://www.mage.space/developers/models/wan-video-v2-2)** — Wan Video v2.2 API: open-weights video with a Lightning variant. From 40 Gems ($0.04) per clip, no subscription, via REST or MCP.
- **[Seed Audio API: Audio Generation](https://www.mage.space/developers/models/seed-audio)** — Seed Audio API: voices, music, and sound effects from one prompt. From 19 Gems ($0.019) per clip, no subscription, via REST or MCP.
- **[MCP Server: Images, Video & Audio in Claude, ChatGPT & Cursor](https://www.mage.space/developers/mcp)** — Remote MCP server for AI image, video, and audio generation. Connect Claude, ChatGPT, Claude Code, Grok, or Cursor to 28 models with OAuth sign-in and 16 tools.
- **[Generate Images & Video in Claude — Mage MCP Connector](https://www.mage.space/developers/mcp/claude)** — Add Mage to Claude as a custom connector and generate images, video, and audio in any chat with Mango 3, Cherry 2 Pro, and more. Sign in with Mage; no API key needed.
- **[Generate Images & Video in ChatGPT — Mage MCP App](https://www.mage.space/developers/mcp/chatgpt)** — Add Mage to ChatGPT as an MCP app and generate images, video, and audio with Mango 3, Cherry 2 Pro, Seed Audio, and more. OAuth sign-in, pay per generation in Gems.
- **[Generate Images & Video in Claude Code — Mage MCP Server](https://www.mage.space/developers/mcp/claude-code)** — Add the Mage MCP server to Claude Code with one command and generate images, video, and audio for your project. Upload local files, sign in with OAuth or an API key.
- **[Generate Images & Video in Grok — Mage MCP Connector](https://www.mage.space/developers/mcp/grok)** — Connect Mage to Grok as a custom MCP connector and generate images, video, and audio with Mango 3, Cherry 2 Pro, and more, paid per generation in Gems.
- **[Generate Images & Video in Cursor — Mage MCP Server](https://www.mage.space/developers/mcp/cursor)** — Install the Mage MCP server in Cursor with one click and let the agent generate images, video, and audio for your app. OAuth with a public client ID, or an API key.
- **[Mage SDKs: TypeScript, Python & ComfyUI](https://www.mage.space/developers/sdks)** — Official, open-source SDKs for the Mage API: @mage-space/sdk for TypeScript, mage-space for Python, and nodes for ComfyUI. Every image, video, and audio model in a few lines.
- **[TypeScript SDK for AI Image, Video & Audio — Mage API](https://www.mage.space/developers/sdks/typescript)** — The official Mage TypeScript SDK (@mage-space/sdk): typed configs for 28 image, video, and audio models, automatic polling, safe retries, and uploads. Open source, pay per generation in Gems.
- **[Python SDK for AI Image, Video & Audio — Mage API](https://www.mage.space/developers/sdks/python)** — The official Mage Python SDK (mage-space): sync and async clients, typed configs for 28 image, video, and audio models, polling, safe retries, and uploads. Open source.
- **[ComfyUI Nodes for Mango, Cherry & Seed Audio — Mage](https://www.mage.space/developers/comfyui)** — Mage for ComfyUI: open-source nodes that run Mango, Cherry, Seed Audio, and 25 more image, video, and audio models in your workflows. No GPU or model downloads; pay per generation in Gems.

---

## 🧩 Mage Skills — Agent Skills for Claude, Codex, Cursor, and Other AI Agents

Mage skills are free, open-source (MIT) Agent Skills that teach an AI agent a complete creative job and run it with Mage's image, video, and audio models: product photoshoots, marketplace listing images, ad variations, resizes, and translations, UGC and faceless videos, YouTube thumbnails, and consistent characters, plus the Mage apps as skills: face, character, and outfit swaps, inpainting, refining, relighting, new camera angles, character design and reference sheets, video edits and swaps, lip sync, music videos, and two-character scenes. They generate through the Mage MCP server (`https://mcp.mage.space/mcp`), so the agent needs the Mage connector; each generation is paid in Gems from the user's Mage account.

- **Browse**: [https://www.mage.space/skills](https://www.mage.space/skills)
- **Install (Claude Code, Codex, Cursor, and other agents)**: `npx skills add mage-space/skills`
- **Claude Code plugin**: `/plugin marketplace add mage-space/skills`, then `/plugin install mage@mage`
- **Source**: [https://github.com/mage-space/skills](https://github.com/mage-space/skills)
- **Connect Mage first**: [https://www.mage.space/mcp](https://www.mage.space/mcp)

### Skills (27)

- **[Generate](https://www.mage.space/skills/generate)** (`/mage-generate`, Essentials) — Any image, video, or audio on the right Mage model, priced before it runs and delivered as a link. Runs on Mango 3, Cherry 2 Pro, Seed Audio, GPT Image 2.5 Flare, Guava 2 Pro.
- **[Consistent Characters](https://www.mage.space/skills/characters)** (`/mage-characters`, Essentials) — Save a character from one portrait, give them a voice, and keep the same face in every image and video. Runs on Mango 3, Cherry 2 Pro, Guava 2 Pro, Seed Audio.
- **[Product Photoshoot](https://www.mage.space/skills/product-photoshoot)** (`/mage-product-photoshoot`, E-commerce) — Studio, lifestyle, hands-on, concept, and Pinterest shots of your product from one phone photo. Runs on Mango 3, Guava 2 Pro, GPT Image 2.5 Flare.
- **[Marketplace Listing Images](https://www.mage.space/skills/marketplace-cards)** (`/mage-marketplace-cards`, E-commerce) — A white-background main image, infographics, and A+ modules for Amazon and other marketplaces. Runs on Mango 3, GPT Image 2.5 Flare, Guava 2 Pro.
- **[Ad Multiplier](https://www.mage.space/skills/ad-multiplier)** (`/mage-ad-multiplier`, Paid Ads) — Turn one ad into test variants: new talent, outfit, setting, or headline, with everything else locked. Runs on Mango 3, GPT Image 2.5 Flare, Cherry 2 Pro.
- **[Ad Resizer](https://www.mage.space/skills/ad-resizer)** (`/mage-ad-resizer`, Paid Ads) — Recompose a static ad for feed, Stories, Reels, and banners without cropping or losing a word. Runs on GPT Image 2.5 Flare, Mango 3.
- **[Ad Localizer](https://www.mage.space/skills/ad-localizer)** (`/mage-ad-localizer`, Paid Ads) — Native-sounding versions of your static ads in other languages, same layout, same offer. Runs on GPT Image 2.5 Flare.
- **[UGC Video](https://www.mage.space/skills/ugc-video)** (`/mage-ugc-video`, UGC) — Creator-style product videos with a scripted hook and real spoken lines, for TikTok, Reels, and ads. Runs on Cherry 2 Pro, Mango 3, Lemon.
- **[Faceless Video](https://www.mage.space/skills/faceless-video)** (`/mage-faceless-video`, Faceless) — Narrated explainers and stories in ten illustrated styles, researched, scripted, and voiced. Runs on Cherry 2 Pro, Mango 3, Seed Audio, Lemon.
- **[YouTube Thumbnail](https://www.mage.space/skills/youtube-thumbnail)** (`/mage-youtube-thumbnail`, Creators) — Truthful, high-click thumbnails with your face kept exact, emotion variants, and headline edits. Runs on Mango 3, GPT Image 2.5 Flare.
- **[Face Swap](https://www.mage.space/skills/face-swap)** (`/mage-face-swap`, Photo Edits) — Put one face and hairstyle on the person in another photo, with the pose, outfit, and scene kept. Runs on Mango 3, Mango 3 Turbo.
- **[Character Swap](https://www.mage.space/skills/character-swap)** (`/mage-character-swap`, Photo Edits) — Replace the whole person in a photo with another character, keeping the scene and the pose. Runs on Mango 3, Mango 3 Turbo.
- **[Outfit Swap](https://www.mage.space/skills/outfit-swap)** (`/mage-outfit-swap`, Photo Edits) — Dress the person in a photo in the clothes from another image, keeping who they are and where they stand. Runs on Mango 3, Mango 3 Turbo.
- **[Inpaint](https://www.mage.space/skills/inpaint)** (`/mage-inpaint`, Photo Edits) — Change, remove, or add one thing in an image and leave every other pixel as it was. Runs on Mango 3, Mango 3 Turbo.
- **[Refine](https://www.mage.space/skills/refine)** (`/mage-refine`, Photo Edits) — Fix eyes, fix hands, or add detail to an image without changing what is in it. Runs on Mango 3, Mango 3 Turbo.
- **[Relight](https://www.mage.space/skills/relight)** (`/mage-relight`, Photo Edits) — New light for a photo: the direction, hardness, brightness, and color you choose, nothing else changed. Runs on Mango 3, Mango 3 Turbo.
- **[Angles](https://www.mage.space/skills/angles)** (`/mage-angles`, Photo Edits) — See a photo from a new camera angle: the same subject, pose, and scene, shot by a second camera. Runs on Mango 3, Mango 3 Turbo.
- **[Recreate](https://www.mage.space/skills/recreate)** (`/mage-recreate`, Photo Edits) — Turn a reference image into a detailed prompt, then make new images in its look. Runs on Mango 3.
- **[Character Builder](https://www.mage.space/skills/character-builder)** (`/mage-character-builder`, Character Design) — Design a new character from a list of traits, then save them as a reusable @handle. Runs on Guava, Mango 3.
- **[Character Muse](https://www.mage.space/skills/character-muse)** (`/mage-character-muse`, Character Design) — Start from a portrait that is close and steer it with face presets or plain words until it is right. Runs on Mango 3, Mango 3 Turbo.
- **[Character Sheet](https://www.mage.space/skills/character-sheet)** (`/mage-character-sheet`, Character Design) — Turn one image of a character into a reference sheet: face, four full-body views, and details. Runs on Mango 3, Mango 3 Turbo.
- **[Video Editor](https://www.mage.space/skills/video-editor)** (`/mage-video-editor`, Video Edits) — Change what a video shows with one instruction, keeping its motion, camera, and timing. Runs on Cherry 2 Pro, Lemon.
- **[Video Face Swap](https://www.mage.space/skills/video-face-swap)** (`/mage-video-face-swap`, Video Edits) — Swap a face into every frame of a clip, keeping the motion, expressions, and lip movements. Runs on Cherry 2 Pro, Lemon.
- **[Video Character Swap](https://www.mage.space/skills/video-character-swap)** (`/mage-video-character-swap`, Video Edits) — Replace the whole person in a clip with your character, body and outfit included. Runs on Cherry 2 Pro, Lemon.
- **[Lipsync](https://www.mage.space/skills/lipsync)** (`/mage-lipsync`, Video Edits) — Make a portrait speak a voice clip, with the lips in sync and the look of the image kept. Runs on Cherry 2 Pro, Seed Audio, Lemon.
- **[Music Video](https://www.mage.space/skills/music-video)** (`/mage-music-video`, Video Edits) — Your characters sing and dance to a track in a music video with cinematic camera work. Runs on Cherry 2 Pro, Seed Audio, Lemon.
- **[Scene Builder](https://www.mage.space/skills/scene-builder)** (`/mage-scene-builder`, Video Edits) — Two characters in a ready-made video scene, from a dinner date to a boxing match, each one kept recognizable. Runs on Cherry 2 Pro, Lemon.

---

## 🏢 About Mage

**[About Mage](https://www.mage.space/about)** — the mission, the team, and what you can create.

- **Founded**: 2022
- **Built by**: Ollano, Inc. — a small, independent team focused on making advanced AI image and video tools accessible to anyone with an idea.
- **Headquarters**: 368 9th Ave, 6th Floor, New York, NY 10001
- **Contact**: [mage@mage.space](mailto:mage@mage.space)
- **Mission**: Launched in 2022, Mage is on a mission to make AI image and video creation accessible to anyone with an idea.

### 📊 Scale

- **10M+** users
- **1B+** creations
- **~300K** images generated per day
- **~30K** videos generated per day

### 🤝 Built on Leading AI Image & Video Models

Mage runs models from leading AI labs and research groups, including:

- OpenAI
- Google
- Black Forest Labs
- ByteDance
- Alibaba
- Wan
- fal
- xAI

### 💬 What People Often Ask

- **Who makes Mage?** Mage is built by Ollano, Inc. — a small, independent team focused on making advanced AI image and video tools accessible to anyone with an idea.
- **How is Mage different from other AI image and video generators?** Mage runs entirely in the browser, supports image and video in one workspace, supports creative freedom within clear lawful limits, keeps every creation private by default, and offers a free quota with no card required. Premium plans add unlimited generation, faster speeds, video, and Mage's exclusive fine-tuned models.
- **What content is allowed on Mage?** Creative freedom within clear, lawful boundaries. Illegal content and prohibited material are strictly not allowed. See the Terms and Conditions for the full content policy.
- **Are my creations private?** Yes. Every creation on Mage is private by default. Nothing appears in public feeds unless you explicitly post it.
- **Can I use Mage creations commercially?** Yes. You can use content created on Mage for commercial purposes. Attribution is appreciated but not required.

### 🔗 Connect with Mage

- **X / Twitter**: [@MageSpace_](https://x.com/MageSpace_)
- **YouTube**: [@MageSpace](https://www.youtube.com/@MageSpace)
- **Instagram**: [@magespace_](https://www.instagram.com/magespace_)
- **LinkedIn**: [Mage on LinkedIn](https://www.linkedin.com/company/magespace)
- **TikTok**: [@magespace_](https://www.tiktok.com/@magespace_)
- **Discord**: [Join the Mage Discord](https://discord.gg/GT9bPgxyFP)
- **Documentation**: [docs.mage.space](https://docs.mage.space)
- **Blog**: [blog.mage.space](https://blog.mage.space)
- **Email**: [mage@mage.space](mailto:mage@mage.space)

---

## 📚 Documentation

**[docs.mage.space](https://docs.mage.space)** is the official Mage documentation: every model with the plan that runs it unlimited and its resolution and duration limits, the apps, Characters and References, membership plans and Gems, usage rights, privacy, and content rules. Its facts are generated from the Mage codebase.

- **Index for AI agents**: [https://docs.mage.space/llms.txt](https://docs.mage.space/llms.txt)
- **Full documentation in one file**: [https://docs.mage.space/llms-full.txt](https://docs.mage.space/llms-full.txt)
- **Any page as Markdown**: append `.md` to a docs URL, for example [https://docs.mage.space/models/overview.md](https://docs.mage.space/models/overview.md)
- **MCP server**: [https://docs.mage.space/mcp](https://docs.mage.space/mcp)
- **Start here**: [Models](https://docs.mage.space/models/overview), [Which model should I use](https://docs.mage.space/models/which-model-should-i-use), [Comparing plans](https://docs.mage.space/plans/comparing-plans), [Apps](https://docs.mage.space/apps/overview)

---

## 🗺️ Site Navigation

### Main Pages

- **[Home](https://www.mage.space)** - Primary generation interface with AI model selection, prompt input, and real-time creation tools
- **[About](https://www.mage.space/about)** - Mission, scale, capabilities, partners, and how Mage keeps your work private by default
- **[Documentation](https://docs.mage.space)** - Official Mage documentation: models, apps, plans and Gems, rights, privacy, and content rules
- **[Blog](https://blog.mage.space)** - Official Mage blog with in-depth articles, tutorials, model updates, and community news
- **[Explore](https://www.mage.space/explore)** - Discover trending AI creations, browse by categories, and find inspiration from the community
- **[Leaderboard](https://www.mage.space/leaderboard)** - Rankings of top creators, most popular concepts, and trending models in the community
- **[Social](https://www.mage.space/social)** - Community feed with user posts, comments, likes, and social interactions
- **[Membership](https://www.mage.space/membership)** - Premium subscription tiers (Basic, Pro, Pro+) with enhanced features and faster generation
- **[Updates](https://www.mage.space/updates)** - Latest platform announcements, new model releases, and feature updates
- **[Import](https://www.mage.space/import)** - Import custom models from CivitAI and other sources for specialized generation

### AI Landing and Major Feature Pages

Dedicated landing pages for Mage's AI generation features and major capabilities.

- **[AI Image Generator](https://www.mage.space/ai-image-generator)** - Turn a written prompt into a finished image, right in your browser. Mage is a free-to-start AI image generator with 160+ models behind one prompt box, from Flux 2 and Stable Diffusion to Mage exclusives like Mango 3 and Guava 2, with nothing to install.
- **[AI Video Enhancer](https://www.mage.space/ai-video-enhancer)** - Soft, low-resolution footage goes in and a sharper clip comes back out. Mage's AI video enhancer upscales video up to 1080p with sharper detail and denoised frames, right in your browser, with no watermark. Free to start.
- **[Anime Image Generator](https://www.mage.space/anime-image-generator)** - Make anime images that actually look like anime. Mage is a free-to-start AI anime image generator running Anima, the Pony and Illustrious families, and 160+ other AI models in your browser, with characters that keep their face from image to image.
- **[Anime Video Generator](https://www.mage.space/anime-video-generator)** - Write a scene, or bring a still to life. Mage is an AI anime video generator with anime-first video models like Melon, characters that stay the same shot to shot, and unlimited video generation from Pro. Free to start, in your browser.
- **[AI Image Editor](https://www.mage.space/ai-image-editor)** - Fix the one thing that came out wrong. Mage is a free-to-start AI image editor: mask an area and describe the change, or hand it the whole image and tell it what to improve. Unlimited inpainting and Refine from $10 a month, right in your browser.
- **[AI Clothes Changer](https://www.mage.space/ai-clothes-changer)** - Put a different outfit on the same photo. Mage is a free-to-start AI clothes changer: bring the look you want as a reference image and it swaps the clothes while the face, the pose and the background stay put. Unlimited swaps from $10 a month, in your browser.
- **[Private AI Image Generator](https://www.mage.space/private-ai-image-generator)** - Create AI images and video that stay yours. Mage is a private-by-default AI image and video generator, free to start right in your browser, with your choice of models and nothing shared publicly unless you decide to.
- **[AI Character Generator](https://www.mage.space/ai-character-generator)** - Create a character once, from a single photo or a prompt, and keep the same face across every image and video. Mage is an AI character generator with a reusable Characters system, no model training, and genuinely unlimited generation. Free to start.
- **[AI Video Generator](https://www.mage.space/ai-video-generator)** - Turn a prompt or a photo into video. Mage is an AI video generator and video maker with text-to-video and image-to-video, exclusive video models, and genuinely unlimited generation. Free to start.
- **[AI Image to Video Generator](https://www.mage.space/ai-image-to-video-generator)** - Turn any image into video. Mage is a free AI image to video generator that animates a photo, a still or a piece of art into moving footage, with unlimited generation on Pro and above. Free to start, right in your browser.
- **[AI Influencer Generator](https://www.mage.space/ai-influencer-generator)** - Build a virtual influencer once, then post a whole feed as the same person. Mage is an AI influencer generator built on a consistent Characters system, with image and video, unlimited generation, and a commercial license. Free to start.

### User Features

- **User Profiles** (/u/[id]) - Individual creator profiles showcasing their generated content and statistics
- **Model Playground** (/play/[id]) - Interactive interface for testing specific AI models and concepts
- **Settings** (/settings?tab=account) - Account preferences, generation settings, and privacy controls
- **Notifications** (/notifications) - System alerts, community interactions, and platform updates

### Apps

Mage offers a suite of focused AI tools — each designed to do one thing really well.

- **[Apps Hub](https://www.mage.space/apps)** - Browse all available Mage apps in one place
- **[Image Enhancer](https://www.mage.space/apps/image-enhancer)** - Upscale and refine any image with AI. Supports up to 6K resolution with enhanced detail and quality.
- **[Video Enhancer](https://www.mage.space/apps/video-enhancer)** - Upscale and refine any video with AI. Supports up to 1080p resolution with enhanced detail.
- **[Video Editor](https://www.mage.space/apps/video-editor)** - Edit any video with a prompt and reference images. Describe the change and it is applied across every frame while preserving the original motion and composition.
- **[Video Face Swap](https://www.mage.space/apps/video-face-swap)** - Swap a face into any video seamlessly. The new identity is applied to every frame while preserving the original pose, motion, expressions, and lip movements.
- **[Video Character Swap](https://www.mage.space/apps/video-character-swap)** - Swap an entire character — face, outfit, and body — into any video while preserving the original motion, camera, and background.
- **[Lipsync](https://www.mage.space/apps/lipsync)** - Make any subject speak an audio clip from a single image. Lip movements sync to the audio with natural expressions, powered by Plum, Cherry, or Blueberry.
- **[Music Video](https://www.mage.space/apps/music-video)** - Make your characters perform a music video to any audio track. Performers sing and dance in sync with the music with cinematic camera work, powered by Plum or Cherry.
- **[Skin Enhancer](https://www.mage.space/apps/skin-enhancer)** - Enhance skin texture and detail in portraits with AI for natural, high-quality results.
- **[Face Swap](https://www.mage.space/apps/face-swap)** - Swap faces between two images seamlessly with AI-powered face detection and blending.
- **[Inpaint](https://www.mage.space/apps/inpaint)** - Edit specific areas of any image with a mask and prompt. Paint over the region you want to change and describe what should replace it.
- **[Character Swap](https://www.mage.space/apps/character-swap)** - Replace an entire character in an image while preserving the scene, powered by AI.
- **[Outfit Swap](https://www.mage.space/apps/outfit-swap)** - Swap clothing, footwear, and accessories onto a subject while preserving the person and scene.
- **[Relight](https://www.mage.space/apps/relight)** - Change the lighting direction, intensity, and mood of any photo with AI. Cinematic lighting in one click.
- **[Character Builder: Attributes](https://www.mage.space/apps/character-builder-attributes)** - Create unique AI characters from preset attributes — appearance, style, mood, and setting — and reuse them across image and video generations.
- **[Character Builder: Muse](https://www.mage.space/apps/character-builder-muse)** - Build one-of-a-kind AI characters by exploring references and starter images.
- **[Character Sheet](https://www.mage.space/apps/character-sheet)** - Turn one image of a character into a full character sheet with a face close-up, front, side, and back views, and outfit details. Save the sheet as a character to keep them consistent across generations.
- **[Recreate](https://www.mage.space/apps/recreate)** - Browse and recreate from a wide range of references.
- **[Refine](https://www.mage.space/apps/refine)** - Make focused improvements to any image with preset prompts or your own refinement instructions.
- **[Scene Builder](https://www.mage.space/apps/scene-builder)** - Build a two-character video scene from character images, presets, and custom prompts.
- **[Angles](https://www.mage.space/apps/angles)** - View any image from a new camera angle. Pick an elevation and orientation to regenerate the same scene from a different viewpoint in one click.

### Legal & Support

- **[Privacy Policy](https://www.mage.space/privacy-policy)** - Comprehensive privacy and data protection information
- **[Terms & Conditions](https://www.mage.space/terms-and-conditions)** - Platform usage guidelines and legal terms

---

## 💎 Membership Tiers & Features

Mage offers flexible membership options to unlock the full potential of AI generation:

### 🆓 **Free Plan** - $0/month
- **Pay-per-generation with Gems** - Generate with premium models and concepts on demand
- **Free trial Gems** - New members receive free Gems to explore premium features
- **Commercial license** - Use your creations for commercial purposes
- **Private generations** - Keep your work private

### ⚡ **Basic Plan** - $10/month *(2x Faster)*
- **Everything in Free**, plus:
- **Unlimited generations** - SD1.5, SDXL, SD3.5, Flux 2, Chroma, Z-Image Turbo, Krea 2, GPT Image 2 & 2.5 (Low)
- **2x faster generation** - Skip the queue with priority processing
- **Verified account** - Community recognition and trust badge
- **Gem access** - Premium currency for special features
- **Base models, LoRAs, and Textual Inversions** - Up to 1 each
- **Inpainting & Refine** - Advanced image editing tools
- **500 Monthly Gems** - Renewed every billing cycle

### 🚀 **Pro Plan** - $30/month *(4x Faster)*
- **Everything in Basic**, plus:
- **Unlimited video generation** - Wan, LTX, Melon (up to 480p, up to 5s), Lemon (up to 3s at 480p), MiniMax-H3 Turbo (up to 10s at 768p, up to 15s at 544p)
- **Unlimited character creation** - Consistent AI characters across images and videos
- **4x faster generation** - Highest priority processing
- **Premium base models** - SD1.5, SDXL, SD3.5, Flux 2, Chroma, Z-Image Turbo, Krea 2, HiDream, Mango, Guava, GPT Image 2 & 2.5 (Low), Anima, Grok Image
- **Multiple LoRAs & Textual Inversions (2x)**
- **Enhance (2K)** - AI-powered image upscaling
- **Hi-Res, Style & Face Reference, ControlNet, Animate**
- **1,500 Monthly Gems**

### 💪 **Pro Plus Plan** - $60/month *(4x Faster)*
- **Everything in Pro**, plus:
- **Premium video models** - Wan, LTX, Cherry, Melon, Plum Max (up to 5s at 480p), Kiwi, Blueberry 2 (up to 4s at 720p, up to 2s at 1080p), Raspberry (up to 4s at 720p, up to 2s at 1080p), Berry (up to 3s at 720p), Lemon (up to 7s at 480p, up to 3s at 720p), MiniMax-H3 Turbo (up to 15s at 768p) and base model (up to 5s at 768p), Grok Video (up to 5s at 480p, up to 3s at 720p)
- **Enhance (4K)** - Ultra high-resolution image upscaling
- **Multiple LoRAs & Textual Inversions (5x)**
- **Import custom models** - SD1.5 and SDXL (Pony, Illustrious) checkpoints, LoRAs, Textual Inversions, and Lycoris; LoRA imports for Wan, Z-Image, Anima, Krea 2, and MiniMax-H3
- **Video LoRA** - Up to 1 custom video model enhancement
- **Queue up to 10 generations at once**
- **3,500 Monthly Gems**

### 👑 **Max Plan** - $200/month *(4x Faster)*
- **Everything in Pro Plus**, plus:
- **Unlimited Motion Control** - Reference video-driven character animation via Pear (5s)
- **More premium video models** - Wan, LTX, Cherry, Cherry 2 Pro (up to 4s at 480p), Melon, Plum (up to 5s at 768p, up to 4s at 2K), Plum Max (up to 9s at 480p, up to 5s at 768p), Kiwi, Blueberry, Raspberry, Berry (up to 4s at 720p), Lemon (up to 11s at 480p, up to 5s at 720p, up to 2s at 1080p), MiniMax-H3 base model (up to 7s at 768p), Grok Video (up to 7s at 480p, up to 5s at 720p)
- **HD video** - 480p with Cherry, Cherry 2 Pro, Grok Video, Lemon, Plum Max; 720p with Kiwi, Blueberry, Raspberry, Berry, Grok Video, Melon, Lemon; 768p with Plum, Plum Max, MiniMax-H3; 1080p with Blueberry 2, Raspberry, Melon, Lemon; 2K with Plum
- **Longer video duration** - 15s at 768p with MiniMax-H3 Turbo; 11s at 480p with Lemon; 10s with Kiwi; 9s at 480p with Plum Max; 7s at 480p with Grok Video; 6s at 720p with Blueberry 2, Raspberry, Melon; 5s with Cherry, Grok Video (720p), and Lemon (720p); 5s at 768p with Plum and Plum Max; 4s at 2K with Plum; 4s at 1080p with Blueberry 2, Raspberry; 4s with Berry; 4s at 480p with Cherry 2 Pro
- **Multiple Video LoRAs (2x)**
- **15,000 Monthly Gems**

### 🎯 **Key Membership Benefits**
- **Speed multipliers** - Skip queues with 2x-4x faster generation
- **Advanced AI models** - Access to latest Stable Diffusion, Flux 2, Krea 2, Mango, and exclusive Mage models
- **Professional tools** - Inpainting, enhancement, ControlNet, and more
- **Video capabilities** - Text-to-video and image-to-video with Cherry, Lemon, MiniMax-H3, Kiwi, Blueberry, Raspberry, Berry, Wan, and more
- **Custom models** - Import and use specialized models from the community
- **Motion Control** - Apply reference video movement to AI characters (Max plan)
- **Commercial licensing** - Use all creations for business purposes

*All plans include commercial licensing and private generation capabilities.*

---

## 📰 Latest Updates & Announcements

> Mage articles and blog posts are hosted at [blog.mage.space](https://blog.mage.space).

### [Unlimited Nano Banana 2.1, More MCP Skills & Other Improvements](https://blog.mage.space/article/unlimited-nano-banana-21-more-mcp-skills-other-improvements/92dd2c93-1322-4c7a-aaaf-7c206a84b444)

*Published: 10/9/2026*

We’re excited to launch Nano Banana 2.1. Now available with Unlimited Generations for Pro Plus and Max members!

---

### [Best Image Generator MCP Servers for Claude (2026)](https://blog.mage.space/article/best-image-generator-mcp-servers-claude/870d2935-871d-4c04-a1c8-5b5831255931)

*Published: 10/8/2026*

Claude can't make images on its own. These 5 MCP servers add image generation to Claude, Claude Desktop and Claude Code, with setup and pricing for each.

---

### [Best AI Image Generators With Advanced Control (2026)](https://blog.mage.space/article/best-ai-image-generators-advanced-control/7111a27b-c2d1-47bf-82a0-70991ae15cba)

*Published: 10/8/2026*

5 browser platforms ranked on ControlNet, LoRA stacking and model imports, with real prices. For power users ready to stop running Stable Diffusion locally.

---

### [5 Best AI Video Generators in 2026, Ranked](https://blog.mage.space/article/best-ai-video-generators-2026/bcf48636-a2e7-4d2b-8136-7f3185a548c4)

*Published: 10/8/2026*

We ranked the 5 best AI video generators of 2026 on realism, prompt response, clip length and audio, from Cherry 2 Pro and Kling 3.0 to Veo 3.1 and Runway Gen-4.5.

---

### [5 Best AI Art Generators (2026) for Concept Art](https://blog.mage.space/article/best-ai-art-generators/4bd724a2-96b8-4661-9f38-11f2674ec868)

*Published: 10/8/2026*

The 5 best AI art generators of 2026 for concept art, painterly styles and illustration, compared on style range, prompt response, pricing and usage rights.

---

### [Mango 3 Turbo, Mage API SDKs & Mage MCP Skils](https://blog.mage.space/article/mango-3-turbo-mage-api-sdks-mage-mcp-skils/18677ac9-fe8f-4112-9ea9-476016db0b35)

*Published: 10/2/2026*

We’re excited to introduce Mango 3 Turbo, a faster and more accessible version of our flagship image model, Mango 3. Now available with Unlimited Generations for Pro, Pro Plus, and Max members. Try it out and let us know what you think!

---

### [Mage API, MCP & Projects](https://blog.mage.space/article/mage-api-mcp-projects/e811882f-25bf-48f5-93be-e00eb790b598)

*Published: 9/26/2026*

Bring Mage into your own apps, workflows, and favorite AI assistants. Our API and MCP are now available in beta, giving you new ways to create with Mage beyond the website.

---

### [The 5 Best AI Tools for Fanvue Creators in 2026](https://blog.mage.space/article/best-ai-tools-fanvue-creators-2026/e85b4b49-d627-4487-9e97-b6284ae206a7)

*Published: 9/25/2026*

The 5 AI tools Fanvue creators actually use in 2026, compared on character consistency, video, output limits and commercial rights.

---

### [AI Influencer Generator: Best Tools for Instagram in 2026](https://blog.mage.space/article/best-ai-image-generators-instagram-influencers-2026/28020468-ac09-4462-bef7-96a353574843)

*Published: 9/25/2026*

Compare the 5 best AI image generators for building a consistent Instagram influencer in 2026, ranked on identity control, realism, and cost at volume.

---

### [How to Use Mage Agent: From Vague Idea to Generation](https://blog.mage.space/article/how-to-use-mage-agent/c8c9415a-15c3-40b9-bb29-e8539a3f19e2)

*Published: 9/25/2026*

Mage Agent turns a vague idea into the generation you want, and helps you pick models and fix prompts along the way. Here's how to use it.

---

### [Best AI Tools for Consistent Anime Characters (2026)](https://blog.mage.space/article/best-ai-tools-consistent-anime-characters-2026/06d84613-d377-459c-81ce-2db6c4943735)

*Published: 9/25/2026*

5 AI tools that keep an anime character's face, hair and outfit the same from image to image, compared on reference locks, LoRA support and price.

---

### [How to Run Stable Diffusion Online in 2026](https://blog.mage.space/article/run-stable-diffusion-online-2026/cef4d0fe-8113-43b4-a5c1-a990049a8a51)

*Published: 9/21/2026*

Run Stable Diffusion 1.5, SDXL, and 3.5 in your browser with no GPU and no setup. How the model works, which version to pick, and what it costs.

---

### [Best AI Image Generators With LoRA Support](https://blog.mage.space/article/best-ai-image-generators-lora-support/ecc26f80-caeb-46ae-a428-0c2ac7483d21)

*Published: 9/21/2026*

Compare 5 AI image generators with LoRA support on architecture coverage, stacking limits and cost per image. See which ones run Flux LoRAs.

---

### [Best AI Image-to-Image Generators in 2026](https://blog.mage.space/article/best-ai-image-to-image-generators-2026/49b73c78-12e5-4a93-aadf-6973535cea1c)

*Published: 9/21/2026*

The 5 best AI image-to-image generators in 2026, compared on mask control, resolution and what a retry costs. Mango 3 leads for character work.

---

### [Mage Agent](https://blog.mage.space/article/mage-agent/caea0fbb-9f44-4a1d-847e-64847a2f1ce6)

*Published: 9/11/2026*

We’re excited to launch Mage Agent — simply tell it what you want to create, and it will handle the rest.

---

### [The 5 Best Platforms for Unlimited GPT Image 2.5 in 2026](https://blog.mage.space/article/best-platforms-gpt-image-2-5-2026/0a79b7b1-14bd-4858-9d42-e27a37e30a97)

*Published: 9/11/2026*

GPT Image 2.5 runs on a dozen platforms, and the billing differs more than the output. Here's where to run it, what each one costs, and what you get.

---

### [Best Online MiniMax H3 Generator in 2026 (No GPU Needed)](https://blog.mage.space/article/best-online-minimax-h3-generator-2026/e65e50e3-d9b0-46c5-9596-fbba06aeaf15)

*Published: 9/11/2026*

Where to run MiniMax H3 online in 2026, compared on unlimited generation, resolution, variants and price. Mage runs H3 unlimited from $30 a month.

---

### [GPT Image 2.5 & Plum Max Video](https://blog.mage.space/article/gpt-image-25-plum-max-video/e7dc9e99-4068-4eaf-9ccd-11a4af52a62f)

*Published: 9/9/2026*

We’re excited to introduce GPT Image 2.5, a state-of-the-art image model — now available with Unlimited Generations for all Premium members.   GPT Image 2.5 Flare: https://www.mage.space/play/gpt-image-25-flare-189b1da81691435aafa926497b56b85f  GPT Image 2.5 Sunburst: https://www.mage.space/play/gpt-image-25-sunburst-b876a82996fc4a00ac9b6ef324a3a106  Or find it via Advanced → Select Model Architecture → GPT Image 2

---

### [MiniMax H3 & Mage Agent (Beta)](https://blog.mage.space/article/minimax-h3-mage-agent-beta/08e5d3cc-6a1f-440d-90e8-05908886047d)

*Published: 9/5/2026*

We are excited to launch MiniMax H3, a leading open-source video model family. H3 is fully compatible with Characters, References, and Character Voices, supports LoRAs, and is now available with Unlimited Generations for Pro, Pro Plus, and Max members.

---

### [The 5 Best AI Video Character Swap Tools in 2026](https://blog.mage.space/article/best-ai-video-character-swap-tools-2026/12641ab0-2977-41f4-87be-78cc67513c8c)

*Published: 9/4/2026*

Some tools replace a character in your footage. Others only animate a character from a driving video. We compare 5 on which job they actually do.

---

## 🎨 Top AI Concepts

*Popular community-created concepts and characters for AI generation*

1. **[AbsoluteDefiant](https://www.mage.space/play/absolutedefiant-ea2cd96f595944ca9a8b6f40a4087c93)** - No description available...
2. **[Z-Image Turbo](https://www.mage.space/play/z-image-turbo-fc43536b26ed41a0b6b0116883f86b26)** - A powerful and lightning-fast new image model that excels in photorealism, human anatomy, and text r...
3. **[General Fast](https://www.mage.space/play/general-fast-49cc148d6b79633a3e1a2b33e77b9f5d)** - No description available...
4. **[Wan Video 2.2 Fast](https://www.mage.space/play/wan-video-22-fast-1671da7abd264bf79b4689d1e793b5b7)** - A faster version of Wan Video 2.2...
5. **[Guava](https://www.mage.space/play/guava-06940b0d70db4a8eb300b0c0b1bbfa58)** - Mage exclusive fruit model great for photorealism. ...
6. **[Guava Pro](https://www.mage.space/play/guava-pro-d0757cf499904d7ea3d281896fdf20f7)** - Mage exclusive fruit model great for photorealism....
7. **[Stable Diffusion 3.5 Turbo](https://www.mage.space/play/stable-diffusion-35-turbo-e95838cfb6f7470ca26b93cbef48fe56)** - No description available...
8. **[AAAbsolute Realism V3](https://www.mage.space/play/aaabsolute-realism-v3-3236082edc7f429a8091bad71d130a7f)** - Mage Exclusive Model for Photorealism...
9. **[AAAbsolute Amateurs V2](https://www.mage.space/play/aaabsolute-amateurs-v2-d3b30289935f4635813f7d087406a3f6)** - Mage Exclusive Model for photorealism. Suggestive. (Fast!)...
10. **[Analog Madness SDXL V2](https://www.mage.space/play/analog-madness-sdxl-v2-88afca9279394fc4b0375c306046b78b)** - No description available...
11. **[Anime](https://www.mage.space/play/anime-efa780b141eb1d629438168533b94961)** - No description available...
12. **[Guava Pro 1.5](https://www.mage.space/play/guava-pro-15-688f095035b54cfa8a006be1842b81e3)** - Mage exclusive fruit model great for photorealism....
13. **[MiniMax-H3 Turbo](https://www.mage.space/play/minimax-h3-turbo-10fc8b4d151f4bda8a6e9e8c07ba3464)** - MiniMax-H3 with the 8-step distillation receipt. Fast video with reference support!...
14. **[GPT Image 2](https://www.mage.space/play/gpt-image-2-ccde44131f954e19accecddac54e1f64)** - A brand new model by OpenAI that excels in prompt understanding, text rendering, editing, and photor...
15. **[Pony Madness V2](https://www.mage.space/play/pony-madness-v2-43b37c0ec02b4be786095ad6f53bf39f)** - No description available...
16. **[Semi-Realistic Portrait](https://www.mage.space/play/semi-realistic-portrait-dcd5fc8698b1995116998a5fcae61ca9)** - No description available...
17. **[Female Manga Characters (Try Any Name)](https://www.mage.space/play/female-manga-characters-try-any-name-ebdd8c6d043e4081bdea1f676fb58aed)** - Create any female manga character with this one!...
18. **[Pony Madness V3](https://www.mage.space/play/pony-madness-v3-ff774254bc9f422d9e9218e7e2960f10)** - No description available...
19. **[Realistic Vision V5](https://www.mage.space/play/realistic-vision-v5-b7cc7e290273b67085b5d08942087e67)** - No description available...
20. **[Lemon](https://www.mage.space/play/lemon-23d5dfa06f664eb782b27ac9ebe089cb)** - Mage's newest exclusive fruit video model, delivering Seedance 2-level quality and greater creative ...
21. **[Stable Diffusion 3.5 Large](https://www.mage.space/play/stable-diffusion-35-large-5e307b1534254f3eb22c36daa1b218bd)** - No description available...
22. **[Photorealism](https://www.mage.space/play/photorealism-fb1d880449b24af4add076b278dc4951)** - No description available...
23. **[RealVisXL V5](https://www.mage.space/play/realvisxl-v5-cd28d1b7f6129e71e3f4e7f649f6771d)** - No description available...
24. **[Pony Realism V2](https://www.mage.space/play/pony-realism-v2-5d67dfff3acc1f0a2923fb320dfcddf2)** - No description available...
25. **[Pony Madness](https://www.mage.space/play/pony-madness-b6ae8b91db7249b8842d79100679c2b1)** - No description available...
26. **[Photonic Fusion SDXL Finale](https://www.mage.space/play/photonic-fusion-sdxl-finale-f2a400da295243f8b2590f6a68a90b7f)** - No description available...
27. **[AAAbsolute Realism](https://www.mage.space/play/aaabsolute-realism-ec9cd32c06b5446ca234b74a69556d54)** - No description available...
28. **[Semi-Realistic Anime](https://www.mage.space/play/semi-realistic-anime-1f4039f7bcc145f6994087d1a4b94d22)** - No description available...
29. **[Prefectious XL V2](https://www.mage.space/play/prefectious-xl-v2-cef6d5bc053c49b3bde8e7ea7eb9df33)** - No description available...
30. **[Beautiful Portrait](https://www.mage.space/play/beautiful-portrait-fedc078d5a66209e8354b8add33561a1)** - No description available...
31. **[Grok Image Quality](https://www.mage.space/play/grok-image-quality-03b401a0324e42c1ab5463debc7b6568)** - A fast, high-quality image model by xAI. This model is fully compatible with Characters and Referenc...
32. **[Anima](https://www.mage.space/play/anima-e0dcd7f75a244b47be4c459662170204)** - Incredible anime-focused model with exceptional artistic abilities. Use Danbooru tags....
33. **[Toon Anime Portrait](https://www.mage.space/play/toon-anime-portrait-ab589a6444864cd6a9bfd94c170f7140)** - No description available...
34. **[AAAbsolute Realism V2](https://www.mage.space/play/aaabsolute-realism-v2-17f2c5712114454f81e52e0045e34c4b)** - Ideal for IG or Onlyfans girls! Focused on amateur photos. (Exclusive & Pony)...
35. **[Ecliptious](https://www.mage.space/play/ecliptious-1afe8c8e493344fa9f029ea1f6309a18)** - No description available...
36. **[Amanatsu Illustrious](https://www.mage.space/play/amanatsu-illustrious-9f140c9cbbbf467788f7a69bc7fd34af)** - No description available...
37. **[AniReal Pony XL](https://www.mage.space/play/anireal-pony-xl-8c08b7556d0342d397b60d10d8dc446e)** - No description available...
38. **[CyberRealistic Pony V11](https://www.mage.space/play/cyberrealistic-pony-v11-0af4a8cfceb84cd1bcf2d10a73ee67bc)** - No description available...
39. **[AbsoluteReality](https://www.mage.space/play/absolutereality-3a412502522242f6c02e7395c9298220)** - No description available...
40. **[DucHaitenPonyReal V2](https://www.mage.space/play/duchaitenponyreal-v2-a1a279ae8cfbbab7ce16a870b2a15034)** - No description available...
41. **[Guava 2 Pro](https://www.mage.space/play/guava-2-pro-ce440d464ee94559bd9bcdb2d36809b5)** - Our newest exclusive fruit image model family, designed for photorealism and greater creative freedo...
42. **[CortadoIllustrious Enhanced](https://www.mage.space/play/cortadoillustrious-enhanced-a2485430545940d0a34eae0e8c57fc18)** - Make any female anime character with this one!...
43. **[CyberRealistic Pony V8.5](https://www.mage.space/play/cyberrealistic-pony-v85-9b09f08d0a0e4ba98a02cf13425f9b44)** - No description available...
44. **[AAAbsolute Amateurs](https://www.mage.space/play/aaabsolute-amateurs-a668223dfc9f4ce4a89a4a2a06293bdd)** - Mage Exclusive Model great for amateur model portrait. Could be suggestive....
45. **[Cinematic Dynamic](https://www.mage.space/play/cinematic-dynamic-891895cac0e6d5fa22aea7d3d77c2e37)** - No description available...
46. **[Animage Pony XL](https://www.mage.space/play/animage-pony-xl-05d04289a2142f35a9cd6d486a9c50d7)** - No description available...
47. **[LotusXL](https://www.mage.space/play/lotusxl-369412a617c8a930c0c669f607c5e07d)** - No description available...
48. **[Photonic Fusion SDXL V2](https://www.mage.space/play/photonic-fusion-sdxl-v2-b0ad3bcb58824e4d80a7055fcf603756)** - No description available...
49. **[Cherry 2 Pro](https://www.mage.space/play/cherry-2-pro-531c7e0c2be14df699a599af606cf7c8)** - The next generation of our most powerful video model. Cinematic and incredible quality! Supports up ...
50. **[Prefect Illustrious XL](https://www.mage.space/play/prefect-illustrious-xl-4f2cb275edc14b2ba8dfcfdfb4b1a674)** - No description available...
51. **[Prefectious XL](https://www.mage.space/play/prefectious-xl-b6f0cd8ab490428e803b2292783d461e)** - No description available...
52. **[CortadoAnime XL V2](https://www.mage.space/play/cortadoanime-xl-v2-118801c126f1416091901848e278b6a5)** - No description available...
53. **[Analog Madness SDXL V3](https://www.mage.space/play/analog-madness-sdxl-v3-76ed63a695aa4428ad24341726d66acf)** - No description available...
54. **[Lucid Dream Pony XL](https://www.mage.space/play/lucid-dream-pony-xl-319e16adef75450cbadc4105c5c0babf)** - No description available...
55. **[AAAbsolute Anime](https://www.mage.space/play/aaabsolute-anime-ccfb1223b0dd46778a347d7bac7ff7df)** - No description available...
56. **[General Fast](https://www.mage.space/play/general-fast-937db6d986233b44d2bdf3ef4ada69b1)** - No description available...
57. **[AniIllustrious](https://www.mage.space/play/aniillustrious-dd633291c4704a7b932596058dc66076)** - No description available...
58. **[Analog Madness SDXL V5](https://www.mage.space/play/analog-madness-sdxl-v5-adf5f8cebb1a4b82a9bd896afaa3a555)** - No description available...
59. **[Mango 3 Turbo](https://www.mage.space/play/mango-3-turbo-91822823d21a403f88397ab31e626866)** - The fastest, most affordable Mango 3 model. Excellent for all categories....
60. **[AAAbsolute Amateurs V4 ](https://www.mage.space/play/aaabsolute-amateurs-v4-a3a5e01217334fd384ba8f452242b8e9)** - Mage exclusive, fast model for amateur portraits. This model is Mature-Native. This is a DMD2 model ...
61. **[Absolute Pony Realism](https://www.mage.space/play/absolute-pony-realism-536934f641ed4859b2c2a74e14f485ba)** - No description available...
62. **[Fast Analog](https://www.mage.space/play/fast-analog-29900c9990e0395b34224dedbc2a0b97)** - No description available...
63. **[DreamUnrealMagexPony](https://www.mage.space/play/dreamunrealmagexpony-c44c7fc83d5c42b0cc8cec1eaf165391)** - No description available...
64. **[AniVerse Pony XL V6](https://www.mage.space/play/aniverse-pony-xl-v6-8673bfcb1121421fa74d7c26d1a600ec)** - No description available...
65. **[Hassaku XL Illustrious](https://www.mage.space/play/hassaku-xl-illustrious-fcc5710642b74ce894e1900a91d1ac3e)** - No description available...
66. **[CyberRealistic](https://www.mage.space/play/cyberrealistic-80698a31bbe963fade119b93ef2b5c67)** - No description available...
67. **[Flux 2 Dev](https://www.mage.space/play/flux-2-dev-323dcb779f924322a35a647ca1a23535)** - The newest version of Flux, excelling in precision and realism....
68. **[AAAbsolute Amateurs V3](https://www.mage.space/play/aaabsolute-amateurs-v3-a6e47378e99c4cefbc18e0b8232b19ab)** - Mage exclusive, fast model for amateur portraits. This model is Mature-Native. This is a DMD2 model ...
69. **[Beautiful Asian Girl](https://www.mage.space/play/beautiful-asian-girl-db04a36761a4659866ffea0f66d48500)** - No description available...
70. **[Realism Reloaded](https://www.mage.space/play/realism-reloaded-79d4cf0c526f4694b7de4d33f8560255)** - No description available...
71. **[AAM XL](https://www.mage.space/play/aam-xl-689aa2c5370488424b82ba53bae40d4c)** - No description available...
72. **[Copax Pony XL V3](https://www.mage.space/play/copax-pony-xl-v3-07177cc94d7ab2f732075a9d0a82d3a9)** - No description available...
73. **[AAAbsolute Furry](https://www.mage.space/play/aaabsolute-furry-ed3890b53eb44344b629d0b5c64baabb)** - Try any furry character names in prompt!...
74. **[Analog Madness SDXL V4](https://www.mage.space/play/analog-madness-sdxl-v4-c643e70ac3544a0eaca073d054080f10)** - No description available...
75. **[Linkavitch Illustrious](https://www.mage.space/play/linkavitch-illustrious-67288c25ed5d43379a6fbca505b6ef95)** - No description available...
76. **[Female Manga Characters (Try Any Name)](https://www.mage.space/play/female-manga-characters-try-any-name-5c887eb0ebc94db1a10434715120c7f2)** - Make any female manga character with this one!...
77. **[AlbedoBase XL V2](https://www.mage.space/play/albedobase-xl-v2-e5c99135af8e889eff7c363039763020)** - No description available...
78. **[SummerLux](https://www.mage.space/play/summerlux-f6e22f659cfa4dd3ae7d033736456981)** - No description available...
79. **[Merged Amateurs](https://www.mage.space/play/merged-amateurs-7bad8d07cf051c2a1d4c708e7c28c815)** - No description available...
80. **[CortadoLatte](https://www.mage.space/play/cortadolatte-84dd80fde590498f93cc825489b9d55a)** - No description available...
81. **[JuggernautXL V11](https://www.mage.space/play/juggernautxl-v11-06ffc7085334774a8a836f59c358e1b3)** - No description available...
82. **[Ghostly Room Portrait](https://www.mage.space/play/ghostly-room-portrait-d49ddff1074840789d2586f6267ccd28)** - No description available...
83. **[Consistent Factor Keter](https://www.mage.space/play/consistent-factor-keter-c08a1da99463d16f3e0019f8a7d6a5a6)** - No description available...
84. **[CyberRealistic V5](https://www.mage.space/play/cyberrealistic-v5-b09d13655b24e8a2160649ac7a1c7be2)** - No description available...
85. **[Prefect Pony XL V3](https://www.mage.space/play/prefect-pony-xl-v3-313a55e9bb3444e2875c0b282c1d7d87)** - No description available...
86. **[Real Life Illustrious Photo](https://www.mage.space/play/real-life-illustrious-photo-c5b80eb5dca844e7a45e708bd95db0b9)** - No description available...
87. **[Realistic Portrait](https://www.mage.space/play/realistic-portrait-2f1eb43493644bdcbba5d837f2e7d83b)** - No description available...
88. **[CFXL V2](https://www.mage.space/play/cfxl-v2-414bd195fbefda1d389c0c538c2cd082)** - No description available...
89. **[Painterly Digital Art](https://www.mage.space/play/painterly-digital-art-b9df731e39034797bc7ec6e4315e2ae2)** - Digital art in an expressionist painterly style....
90. **[Analog Madness SDXL](https://www.mage.space/play/analog-madness-sdxl-6dcae8e0ba3660846398566de7b70918)** - No description available...
91. **[CortadoIllustrious Enhanced V6](https://www.mage.space/play/cortadoillustrious-enhanced-v6-a024be2092bf4781ae141923f121d162)** - Make any female anime characters you desire. (Exclusive & Illustrious)...
92. **[Z-Image](https://www.mage.space/play/z-image-b133f9e38b984358a9a6422c8fe6090f)** - Z-Image Base provides good quality, robust generative diversity, broad stylistic coverage, and preci...
93. **[Amateur Photo Portrait](https://www.mage.space/play/amateur-photo-portrait-df1d72c555d148418a9fba8fad87d6b6)** - No description available...
94. **[Crystal Clear XL One](https://www.mage.space/play/crystal-clear-xl-one-4c8391ee974dff48c8b9f04bb33dc1cf)** - No description available...
95. **[AutumnLux](https://www.mage.space/play/autumnlux-87fac931e93409dd3bc5433b2dc849e3)** - No description available...
96. **[Photonic Fusion V2](https://www.mage.space/play/photonic-fusion-v2-c7ffc978bf12db29a2d27627260f4a01)** - No description available...
97. **[Elegant Toon Portrait](https://www.mage.space/play/elegant-toon-portrait-1245ee01d769a248382ac6dabd2f6670)** - No description available...
98. **[Yami Anime Mix](https://www.mage.space/play/yami-anime-mix-f8e4b7c27136425da209e4fbe908459a)** - Mage exclusive illustrious model great for dark anime....
99. **[AniVerse Pony XL V3](https://www.mage.space/play/aniverse-pony-xl-v3-7f1e1dc38ca541a190665638f5eac0a5)** - No description available...
100. **[DucHaitenPonyXL NoScore V7](https://www.mage.space/play/duchaitenponyxl-noscore-v7-ad7dc294631a4cbda4244e4605d7b3d2)** - No description available...

---

## 🔧 Available AI Models & Architectures

### Image Generation Models
- **Flux 2** - Latest state-of-the-art image generation with exceptional quality and prompt adherence
- **Krea 2** - Krea 2 Turbo, Krea's latest image model, with 1K and 2K resolution and LoRA support; 1K included in unlimited image generation for all members (2K requires Pro)
- **Mango** - Mage-exclusive image model similar to Seedream, optimized for fast and very high-quality generation. **Mango 3 is live now**, alongside Mango 3 Turbo (the fastest and cheapest) and Mango 3S — the latest generation of the Mango family
- **GPT Image 2** - OpenAI's GPT Image 2 alongside GPT Image 2.5 Flare (the faster 2.5 model) and GPT Image 2.5 Sunburst (the premium 2.5 model for precise edits), with Low/Medium/High quality tiers plus Extra High and Max on the 2.5 models, 1K/2K resolution, and support for up to 15 reference images, characters, and semantic references
- **Stable Diffusion XL** - High-resolution image generation with excellent artistic capabilities
- **Stable Diffusion 3.5 Large** - Advanced model with improved text rendering and composition
- **Stable Diffusion 1.5** - Classic model with extensive fine-tune ecosystem
- **Anima** - Anime / illustration / art-focused model with Danbooru-style tag and natural-language prompting, 1K and 2K output, included in unlimited image generation for Pro members and above
- **Grok Image** - xAI's Grok Imagine image model with BASIC, QUALITY, and 2.0 variants (2.0 adds a low/medium quality setting), 1K and 2K resolution, characters and references support, and up to 3 reference images; strict content filter; included in unlimited image generation for Pro members and above

### Video Generation Models
- **Cherry** - Mage-exclusive video model similar to Seedance V2.0, with text-to-video and image-to-video capabilities, first and last frame support, multiple reference images (up to 8), and durations up to 15 seconds. **Cherry 2 Pro is live now** — the newest generation of the Cherry family, similar in quality to Seedance 2.5, with 480p, 720p, and 1080p output and support for characters, reference images, and audio references; exclusive to Cherry 2 Pro: extended durations of 20, 25, and 30 seconds and up to 29 total reference images; runs with Gems for everyone and is included in unlimited video generation for Max (up to 4s at 480p)
- **Kiwi** - Mage-exclusive video model similar to Wan 2.5, with text-to-video and image-to-video capabilities
- **Blueberry** - Mage-exclusive video model similar to Wan 2.6, with advanced motion and consistency
- **Raspberry** - Mage-exclusive text/reference-to-video model with multi-image and character reference support for consistent video generation
- **Berry** - Mage-exclusive state-of-the-art reference-driven video model with output quality comparable to Cherry but with significantly more creative freedom; text-to-video and image-to-video with first frame support, and up to 9 reference images at 720p or 1080p
- **Melon** - Mage-exclusive video model offered as Melon and Melon Pro; supports text-to-video, image-to-video, and start-end-to-video with first and last frame support, plus reference-to-video with characters and up to 7 reference images; optional generated audio, 540p/720p/1080p output, and 3–16s duration
- **Plum** - Mage-exclusive video model similar to MiniMax-H3, offered as Plum and Plum Max; supports text-to-video, image-to-video with first and/or last frame (on Plum Max a last frame must be paired with a first frame), and reference-to-video with characters, up to 9 reference images, and up to 3 audio references for voices and sound; every video is generated with audio (speech, sound effects, ambience); Plum renders 768p/2K output at 4–15s, Plum Max is the faster model at 480p/768p and 5–15s; included in unlimited video generation from Pro Plus (Plum Max up to 5s at 480p) and Max (Plum up to 5s at 768p and 4s at 2K, Plum Max up to 9s at 480p and 5s at 768p); everything else runs on Gems
- **Lemon** - Mage-exclusive all-in-one video model; one model covers text-to-video, image-to-video with a first and/or last frame, and reference-to-video that combines characters, up to 10 reference images, up to 5 reference videos, and up to 5 audio references for voices and sound in a single request; optional generated audio track and seed control; 480p/720p/1080p output and 2–30s duration; included in unlimited video generation from Pro (up to 3s at 480p), Pro Plus (up to 7s at 480p, up to 3s at 720p), and Max (up to 11s at 480p, up to 5s at 720p, up to 2s at 1080p); everything else runs on Gems, and Gems runs take a faster lane with its own gem price
- **MiniMax-H3** (MiniMax H3) - Open-weights 33B video model from MiniMax, hosted on Mage's own GPUs, that generates video and stereo audio together (speech, sound effects, ambience); text-to-video, image-to-video with a first and/or last frame, and reference-to-video with characters, up to 9 reference images, up to 3 reference videos, and up to 3 audio references for voices and sound; 480p/544p/720p/768p output at 24 fps and 5–15s duration (the base model up to 8s); references share a 21-second budget with the clip (each image counts as 2 seconds, each reference video as its own length); MiniMax-H3 Turbo (8-step) and the 50-step MiniMax-H3 base model; included in unlimited video generation from Pro (Turbo up to 10s at 768p, up to 15s at 544p, plus the base model up to 6s at 544p), Pro Plus (Turbo up to 15s at 768p, base model up to 5s at 768p), and Max (base model up to 7s at 768p); everything else runs on Gems
- **Wan 2.2** - Advanced text-to-video and image-to-video capabilities
- **LTX Video** - Fast, efficient video generation for quick iterations
- **Grok Video** - xAI's Grok Imagine video model with 480p and 720p resolution, 1-15s duration, characters and references support, and up to 5 reference images; strict content filter; included in unlimited video generation for Pro Plus (up to 5s at 480p / 3s at 720p) and Max (up to 7s at 480p / 5s at 720p)

### 🔊 Audio Generation Models (NEW)
- **Seed Audio** - Mage's new audio generation architecture. Generate audio from text prompts — clips from 1 to 120 seconds, MP3 output at 48kHz, up to 3 audio references plus an image reference, and fine-grained speech rate, loudness, and pitch controls. Unlimited generation on the standard lane, with fast mode available via Gems

### Enhancement & Utility Models
- **Enhance** - AI-powered image upscaling and quality improvement
- **Refine** - Image refinement and detail enhancement
- **Chroma** - Background removal and replacement

### Advanced Features
- **LoRA Support** - Layer-based model adaptations for specific styles
- **ControlNet** - Precise control over generation composition
- **Textual Inversion** - Custom concept embedding
- **Inpainting** - Advanced image editing capabilities

---

*Generated: 2026-10-14T04:54:23.349Z*

> **Mage** - Where imagination meets artificial intelligence. Create without limits.