Text-to-Image Models

Overview

The Text-to-Image API provides access to 50+ cutting-edge models that generate images from text descriptions. From photorealistic renders to artistic styles, these models cover a wide range of creative use cases.

Text-to-image models are automatically selected when no reference image is provided in the Image Studio.

Model Categories

Flux Models

High-quality photorealistic generation with fast inference times.

Flux Dev — Developer-grade quality with balanced speed
Flux Schnell — Ultra-fast generation for rapid prototyping
Flux Dev Lora — Custom LoRA support for style transfer
Flux Kontext Dev/Pro/Max — Context-aware generation
Flux 2 Dev/Flex/Pro — Next-generation models with enhanced detail
Flux 2 Klein 4b/9b — Lightweight models for cute and clean illustration styles
Flux Pulid — Identity-preserving generation
Flux Redux — Reimagine and transform existing compositions
Flux Krea Dev — Cinematic realism and shallow depth of field

Proprietary Models

Industry-leading commercial models with exceptional quality.

Midjourney v7 — Cinematic composition with variety, stylization, and weirdness controls
Google Imagen4 (Standard/Fast/Ultra) — Photorealistic quality from Google
GPT-4o Text-to-Image — OpenAI’s vision model for diagram and illustration generation
Ideogram v3 — Text rendering and typography specialist
Leonardo AI Phoenix 1.0 & Lucid Origin — Fantasy and concept art
Grok Imagine — X.AI’s creative generation model

Open Source Models

High-performance open models for production use.

SDXL Image — Stable Diffusion XL base model
Hunyuan Image 2.1 & 3.0 — Tencent’s multilingual models
Qwen Image & 2512 — Alibaba’s vision models
Chroma Image — Vibrant chromatic generation
Hidream I1 Fast/Dev/Full — Cartoon and illustration styles
Z Image Turbo/Base — Fast and flexible generation
Perfect Pony XL — Specialized for equine subjects
Neta Lumina — Anime and character art

Specialized Models

Domain-specific models for niche use cases.

Nano Banana & Nano Banana 2 — Multi-resolution with Google Search enhancement
Nano Banana Pro — Professional-grade with 1K/2K/4K resolution options
Seedream 5.0 — ByteDance’s latest with quality presets
Bytedance Seedream v3/v4/v4.5 — Dreamlike aesthetics
Wan2.1/2.5/2.6 — Realistic skin textures and portrait generation
AI Anime Generator — Specialized anime and manga styles
Kling O1 & Vidu Q2 — High-resolution cinematic stills
Reve Text-to-Image — Photorealistic fantasy portraits

Popular Models

Model	Resolution	Aspect Ratios	Speed	Best For
Nano Banana 2	1K/2K/4K	15+ ratios + auto	Fast	Production work with Google Search enhancement
Seedream 5.0	Up to 4K	8 ratios	Fast	High-quality stylized images
Flux Dev	Up to 2048px	Flexible (width/height)	Medium	Photorealistic generation
Midjourney v7	Variable	11 ratios	Slow/Fast/Turbo	Cinematic and artistic compositions
Google Imagen4	Variable	5 ratios	Medium	Google-grade photorealism
Ideogram v3	Variable	5 ratios	Turbo/Balanced/Quality	Typography and text rendering
Flux 2 Pro	Up to 1536px	7 ratios	Medium	Next-gen detail and coherence
Kling O1	1K/2K	8 ratios	Medium	Cinematic ultra-detailed renders

Newly Added Models

These models were recently added to the platform and offer cutting-edge capabilities.

Nano Banana 2

Family: Nano
Resolution: 1K / 2K / 4K
Aspect Ratios: 1:1, 1:4, 1:8, 2:3, 3:2, 3:4, 4:1, 4:3, 4:5, 5:4, 8:1, 9:16, 16:9, 21:9, auto

Features

Powered by Google Gemini 3.1 Flash Image
Google Search enhancement for prompt enrichment
Auto aspect ratio detection
Multiple output formats (JPG, PNG)
Resolution scaling from 1K to 4K

Example Request

{
  "model": "nano-banana-2",
  "prompt": "A futuristic cityscape with glowing neon lights reflected in rain-soaked streets, ultra-detailed 4K photography.",
  "aspect_ratio": "auto",
  "resolution": "4k",
  "google_search": true
}

Seedream 5.0

Family: Seedream
Quality: Basic / High
Aspect Ratios: 1:1, 16:9, 9:16, 4:3, 3:4, 2:3, 3:2, 21:9

Features

ByteDance’s latest generation model
Quality presets for basic and high-fidelity outputs
Wide aspect ratio support
Natural language style descriptions

Example Request

{
  "model": "seedream-5.0",
  "prompt": "A futuristic city with soaring crystalline towers, suspended gardens, and neon-lit skyways under a twin-moon sky, captured in a cinematic, high-detail digital art style.",
  "aspect_ratio": "16:9",
  "quality": "high"
}

Common Input Parameters

Prompt

Type: string
Required: Yes
Max Length: 2–3000 characters (varies by model) Text description of the image you want to generate. The more detailed and specific, the better the results.

For best results, include:

Subject description
Style or artistic direction
Lighting and atmosphere
Composition details
Technical specifications (e.g., “8K”, “cinematic”, “photorealistic”)

Aspect Ratio

Type: string
Options: Varies by model (typically 1:1, 16:9, 9:16, 4:3, 3:4, 21:9, etc.) Defines the width-to-height ratio of the generated image. Some models support 15+ ratios, including ultra-wide (21:9) and auto-detection.

Resolution / Quality

Some models offer resolution presets:

Resolution-based: 1k, 2k, 4k, 480p, 720p, 1080p
Quality-based: basic, high, medium, low, turbo, balanced, quality
Dimension-based: Width/Height in pixels (must be divisible by 64 for Flux models)

Higher resolutions and quality settings consume more credits and take longer to generate.

Number of Images

Type: integer
Range: 1–4 (varies by model)
Default: 1 Generate multiple variations in a single request. Each image is charged separately.

Model-Specific Parameters

Midjourney v7 — Advanced Controls

Speed: relaxed | fast | turbo
Controls generation priority and speed.Variety: 0–100 (step: 5)
Higher values create more diverse results; lower values create consistency.Stylization: 0–1000
Intensity of artistic style. Higher = more stylized; lower = more realistic.Weirdness: 0–3000
Creativity and uniqueness. Higher = unusual; lower = conventional.

Flux Dev Lora — Custom Style Transfer

Model ID: Array of LoRA models from Civitai

{
  "model_id": [
    {
      "model": "civitai:119351@317153",
      "weight": 1.0
    }
  ]
}

Each LoRA can have a weight between 0–4. Maximum 4 models per request.

Ideogram v3 — Typography Specialist

Google Imagen4 — Variants

Imagen4 — Standard quality
Imagen4 Fast — Faster inference
Imagen4 Ultra — Maximum quality (longer generation)

Example Prompts

Photorealistic Portrait

A young woman with freckles and natural makeup, standing in soft sunlight, 
sharp focus, DSLR photo style, ultra-realistic skin texture, 8K detail.

Recommended Models: Flux Dev, Google Imagen4, Wan2.5

Cinematic Scene

A sprawling futuristic city at dusk, illuminated with vibrant neon signs, 
layered skyscrapers, elevated highways with flying cars, warm atmospheric glow, 
ultra-detailed sci-fi architecture, cinematic composition, high contrast, 8K.

Recommended Models: Midjourney v7, Kling O1, Flux 2 Pro

Anime Character

A cheerful anime girl with short pink hair and green eyes, wearing a school uniform, 
standing under cherry blossom trees, soft lighting, anime style, digital painting.

Recommended Models: AI Anime Generator, Neta Lumina, Hidream I1 Dev

Fantasy Concept Art

A majestic elven queen standing in a glowing forest, wearing intricate golden armor 
with emerald details, sunlight rays filtering through the trees, ultra-detailed 
fantasy concept art, cinematic lighting, 8K render.

Recommended Models: Leonardo AI Phoenix, Midjourney v7, Flux 2 Pro

API Endpoints

All text-to-image models use the unified endpoint pattern:

POST /api/v1/{model-endpoint}

Refer to the Generate Image guide for integration details.

Model Selection Tips

For production work: Use Nano Banana 2, Flux Dev, or Seedream 5.0 for reliable quality and speed.For creative exploration: Try Midjourney v7, Leonardo AI models, or Bytedance Seedream variants.For anime/illustration: Use AI Anime Generator, Neta Lumina, or Hidream models.For typography and text: Use Ideogram v3 with Design or General style.For maximum quality: Use Google Imagen4 Ultra, Flux 2 Pro, or Kling O1 at 2K/4K resolution.

Resolution Guide

Use Case	Recommended Resolution	Models
Social media posts	1K (1024×1024)	Most models
Desktop wallpapers	2K (2048×1152)	Nano Banana 2, Flux 2, Kling O1
Print quality	4K (4096×2304)	Nano Banana 2, Seedream 5.0, Bytedance v4.5
Web thumbnails	480p–720p	Google Imagen4 Fast, Flux Schnell

Rate Limits

Each model has different credit costs based on resolution and quality through the Muapi.ai API. Refer to the Muapi.ai documentation for detailed pricing information.

Generating multiple images in a single request (using num_images parameter) charges separately for each image.

API Client

Models

Overview

Model Categories

Popular Models

Newly Added Models

Nano Banana 2

Seedream 5.0

Common Input Parameters

Prompt

Aspect Ratio

Resolution / Quality

Number of Images

Model-Specific Parameters

Example Prompts

API Endpoints

Model Selection Tips

Resolution Guide

Rate Limits

Build docs developers (and LLMs) love

API Client

Models

​Overview

​Model Categories

​Popular Models

​Newly Added Models

​Nano Banana 2

​Seedream 5.0

​Common Input Parameters

​Prompt

​Aspect Ratio

​Resolution / Quality

​Number of Images

​Model-Specific Parameters

​Example Prompts

​API Endpoints

​Model Selection Tips

​Resolution Guide

​Rate Limits

Build docs developers (and LLMs) love

Overview

Model Categories

Popular Models

Newly Added Models

Nano Banana 2

Seedream 5.0

Common Input Parameters

Prompt

Aspect Ratio

Resolution / Quality

Number of Images

Model-Specific Parameters

Example Prompts

API Endpoints

Model Selection Tips

Resolution Guide

Rate Limits