AI 图像生成器
Muapi 为主流文生图模型提供统一的 HTTP 接口。提交提示词、获取请求 ID 并轮询结果，无需为每个模型维护单独的 SDK 或账户。所有模型均按次使用透明计费。
- →通过 POST /api/v1/{model} 访问 Flux、Seedream、GPT-Image、HiDream、Qwen、Reve 和 Midjourney 等模型
- →所有模型使用统一的提交后轮询流程，一次集成即可切换模型
- →Playground 自动生成输入表单，并通过 OpenAPI 提供完整 Pydantic 参数结构
- →仅在生成成功后扣除积分，失败请求不收费
- →每次请求都可以选择 webhook 回调，免去轮询
快速开始
此类别中的所有模型都使用相同的提交后轮询 API。将 flux-schnell 替换为下方列表中的任意模型端点。
# 1. Submit
curl -X POST https://api.muapi.ai/api/v1/flux-schnell \
-H "x-api-key: $MUAPI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"prompt":"a serene mountain lake at dawn, photorealistic"}'
# → {"request_id":"abc123","status":"processing"}
# 2. Poll until completed
curl https://api.muapi.ai/api/v1/predictions/abc123/result \
-H "x-api-key: $MUAPI_API_KEY"排名前 5 的AI 图像生成器模型
| 模型 | 提供商 | 成本 | 适用场景 |
|---|---|---|---|
| gpt-image-2-text-to-image | — | $0.090 | Generate high-quality images from text prompts using GPT Image 2, supporting up to 20,000 character prompts for detailed and precise image creation. |
| flux-dev | — | $0.015 | Generate stunning visuals from simple text prompts. Flux Dev transforms your ideas into high-quality, creative images using powerful AI vision models. Perfect for design, storytelling, concept art, and marketing. |
| flux-schnell | — | $0.003 | Flux Schnell is a lightning-fast image generation model designed for rapid iterations. It delivers good visual quality from text prompts almost instantly, making it perfect for real-time concept testing, brainstorming, and UI-integrated experiences. |
| nano-banana | — | $0.030 | Nano Banana is an advanced AI model excelling in natural language-driven image generation and editing. It produces hyper-realistic, physics-aware visuals with seamless style transformations. |
| nano-banana-pro | — | $0.120 | Nano Banana 2 is the next-generation image generation developed by Google DeepMind, following the original Nano Banana (also known as Gemini 2.5 Flash Image). It offers advanced text-to-image capabilitie with improved resolution. |
全部 70 个模型
krea-v2-turbo
Krea 2 Turbo generates aesthetic-first images natively at 1K to 2K resolution in seconds, supporting both text-to-image and image-to-image workflows.
ai-anime-generator
Create stunning anime-style artwork instantly with our AI Anime Generator. Customize characters, scenes, and styles effortlessly in seconds!
leonardoai-lucid-origin
Lucid Origin is LeonardoAI’s advanced image generation model, designed for ultra-realistic, vibrant, and highly detailed visuals. It excels at creating photorealistic portraits, landscapes, product shots, and stylized art while faithfully following complex prompts.
z-image-base
Z-Image Base is a general-purpose text-to-image model designed for reliable, high-quality image generation from natural language prompts. It focuses on clear composition, good prompt adherence, and versatile output across everyday scenes, product-style visuals, characters, and creative concepts.
wan2.1-text-to-image
WAN 2.1 is a powerful AI model that transforms text prompts into high-resolution, photorealistic images. It excels at detailed object rendering, realistic lighting, and fine textures, making it ideal for visual content, concept art, advertising, and digital storytelling.
flux-2-klein-4b-turbo
Flux-2-Klein-4B Turbo is an ultra-fast, high-efficiency text-to-image model. It is a distilled version of the Klein 4B model, designed for near-instant rendering while maintaining impressive adherence to prompts. Perfect for rapid prototyping, real-time creative tools, and applications where speed is paramount.
wan2.7-text-to-image
Alibaba WAN 2.7 Text-to-Image generates high-quality images from text prompts with thinking mode for enhanced image quality.
hunyuan-image-3.0
Hunyuan Image 3.0 brings together powerful architecture (Mixture-of-Experts + autoregressive style) to produce richly detailed and coherent images from complex prompts. It can read narrative descriptions, render text and signage cleanly, and support multiple visual styles — from photorealism to illustrations.
bytedance-seedream-v4
Seedream v4 generates stunning, high-fidelity images from text prompts. It’s designed for creativity with strong support for realism, fantasy, and artistic styles.
ideogram-v3-t2i
Ideogram v3 is an advanced text-to-image model designed for creating highly detailed and visually striking images directly from text prompts. It’s especially good for artistic compositions, design mockups, concept art, and photorealistic scenes. With strong support for text rendering inside images, it’s widely used for posters, typography-based art, and creative branding.
wan2.7-text-to-image-pro
Alibaba WAN 2.7 Text-to-Image Pro generates high-quality images up to 4K from text prompts with thinking mode for enhanced image quality.
hidream-i1-full
The most advanced version of HiDream I1, delivering high-resolution, detailed images with superior prompt understanding. Best suited for production, content creation, and high-fidelity applications.
flux-kontext-pro-t2i
Flux Kontext Pro T2I offers fast and reliable generation with creative flexibility. It supports stylized prompts, character design, and fantasy themes while maintaining clear subject coherence.
flux-kontext-max-t2i
Flux Kontext Max T2I delivers photorealistic or cinematic-quality images with exceptional detail. It's optimized for high-end visuals — from realistic humans to polished product renders.
gpt4o-text-to-image
Generate images from text prompts using GPT-4o's vision capabilities. Ideal for basic concept visuals, diagrams, and abstract compositions.
flux-schnell
Flux Schnell is a lightning-fast image generation model designed for rapid iterations. It delivers good visual quality from text prompts almost instantly, making it perfect for real-time concept testing, brainstorming, and UI-integrated experiences.
bytedance-seedream-v3
Seedream is designed for generating visually rich and artistic images from text prompts. It excels at fantasy, anime, surrealism, and vibrant color compositions — ideal for creative visuals, storyboards, and concept art.
qwen-image
Generate high-quality, detailed images from text prompts in various styles — from realistic to artistic — perfect for creative visuals, product shots, and concept art.
nano-banana
Nano Banana is an advanced AI model excelling in natural language-driven image generation and editing. It produces hyper-realistic, physics-aware visuals with seamless style transformations.
google-imagen4
Google Imagen 4 is the latest text-to-image AI model from DeepMind, designed to produce stunningly photorealistic images with crisp detail, accurate text rendering, and creative flexibility. It supports high-resolution output (up to 2K), generates visuals in seconds, and embeds SynthID watermarks for authenticity.
google-imagen4-fast
Imagen 4 Fast is optimized for speed and accessibility, allowing you to generate high-quality images in seconds. While slightly less detailed than the Ultra version, it excels at rapid ideation, drafts, storyboarding, and casual creativity.
google-imagen4-ultra
Imagen 4 Ultra is Google’s flagship model, designed for photorealism, rich textures, and production-level imagery. It produces crisp, high-resolution visuals with advanced detail, lighting precision, and natural compositions.
hunyuan-image-2.1
Hunyuan Image is a powerful text-to-image generation model that produces photorealistic and highly detailed visuals. It excels at creating portraits, environments, and concept art with strong consistency and realism. Designed for versatility, it supports both natural photography styles and imaginative artistic outputs.
chroma-image
Croma Image is an advanced text-to-image generation model designed for high-quality, creative, and versatile visuals. It can produce anything from photorealistic portraits and products to imaginative concept art, fantasy illustrations, and cinematic scenes.
perfect-pony-xl
Pony XL is a high-quality image generation model based on Stable Diffusion XL architecture. It specializes in character art, hybrid styles, and producing detailed, polished visuals even with simpler prompts.
wan2.5-text-to-image
WAN 2.5 Text-to-Image generates high-quality, realistic or stylized images from textual descriptions. It supports detailed visual storytelling, cinematic compositions, and versatile styles — from portraits and product shots to landscapes and fantasy scenes.
gpt-image-1.5
GPT-Image-1.5 is a high-quality text-to-image generation model designed for rich visual reasoning, detailed compositions, and strong prompt understanding. It excels at complex scenes, symbolic imagery, cinematic lighting, surreal concepts, product visuals, and imaginative world-building while maintaining coherence and fine detail.
flux-2-klein-9b
Flux-2-Klein-9B is a mid-size text-to-image model that balances detail quality and generation speed. It handles richer lighting, better textures, and more nuanced scenes than smaller variants, while still working well with clear, grounded prompts. Ideal for polished illustrations, product visuals, mascots, and everyday scenes with character.
z-image-turbo
Z-Image Turbo is a high-speed text-to-image model optimized for fast creative generation. It produces detailed, high-contrast, high-resolution images with strong stylization control. Ideal for rapid concept creation, visual exploration, product ideas, fantasy scenes, and cinematic composition tests. Designed for low latency and strong prompt adherence.
flux-2-flex
Flux-2-Flex Text-to-Image is a flexible, high-fidelity generative model capable of producing detailed, imaginative, and stylistically rich scenes from text alone. It excels at surreal concepts, fantasy environments, sci-fi structures, cinematic atmospheres, and high-resolution artistic compositions with strong prompt adherence.
nano-banana-2
Nano Banana 2 (Gemini 3.1 Flash Image) is Google's most advanced image generation model, combining speed with high-fidelity 4K output and revolutionary character consistency.
z-image-p
Z-Image P is based on PiAPI's Qubico/z-image text-to-image model.
qwen-image-2.0
Qwen 2.0 Text to Image model with enhanced realism.
qwen-image-2.0-pro
Qwen 2.0 Pro Text to Image model with maximum realism and fidelity.
tiktok-carousel
AI TikTok Carousel Generator — create viral TikTok carousel posts from a single text prompt. Choose a proven storytelling format (Problem-Solution, Listicle, Tutorial, Before & After), set your slide count (3-10), and get stunning AI-generated images at 1080x1920 portrait resolution, ready to upload to TikTok.
gpt-image-2-text-to-image
Generate high-quality images from text prompts using GPT Image 2, supporting up to 20,000 character prompts for detailed and precise image creation.
flux-3-text-to-image
FLUX 3 Text-to-Image is Black Forest Labs' next-generation multimodal frontier model, jointly trained across image, video, and audio for outputs that are truer to life in every style. It generates highly photorealistic and stylistically flexible images from text prompts, with sharper detail, more coherent composition, and stronger prompt adherence than the FLUX.2 generation.
flux-2-klein-9b-turbo
Flux-2-Klein-9B Turbo is a high-performance, mid-size text-to-image model. This distilled variant of Klein 9B provides a superior balance of speed and detail, delivering richer textures and complex scenes with significantly reduced generation times. Ideal for polished illustrations and character-rich visuals where performance is key.
muse-image-text-to-image
Meta Muse Image Text-to-Image generates images directly from a text prompt, covering subject, scene, composition, lighting, mood, and style in a single pass.
kling-o3-image
Generate detailed photoreal and stylised images from text prompts using Kling O3. Supports 1K/2K/4K resolutions, multiple aspect ratios, and up to 9 outputs per request.
nano-banana-2-lite
Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) is Google's fastest and most cost-efficient text-to-image model, delivering 4-second generation with exceptional prompt adherence, character consistency, and legible in-image text rendering.
bytedance-seedream-5.0-pro
Seedream 5.0 Pro is ByteDance's flagship next-generation text-to-image model, extending Seedream 5.0 Lite with higher-fidelity rendering, deeper visual reasoning, and finer typography control. It targets up to 2K output with stronger scene composition, more consistent multi-character generation, and richer prompt adherence for professional creative workflows.
flux-dev
Generate stunning visuals from simple text prompts. Flux Dev transforms your ideas into high-quality, creative images using powerful AI vision models. Perfect for design, storytelling, concept art, and marketing.
wan3.0-text-to-image
Wan 3.0 Text to Image is an upcoming AI image model. Confirmed API controls, output specifications, and pricing will be published at launch.
sdxl-image
SDXL is a high-quality, large Stable Diffusion model for creating photorealistic and stylized images from text. It excels at fine detail, realistic lighting, and complex scenes.
flux-krea-dev
Flux Krea Dev is a text-to-image model built by Black Forest Labs in collaboration with Krea AI, designed to generate highly photorealistic images that avoid the common 'AI look' artifacts (plastic skin, overexposed lighting, synthetic textures). It emphasizes real texture, natural lighting, and aesthetic control.
neta-lumina
Neta Lumina is a powerful anime-style text-to-image model developed by Neta.art Lab. It’s built on Lumina-Image-2.0, fine-tuned with over 13 million high-quality anime images. It offers strong understanding of multilingual prompts, excellent detail fidelity, support for Danbooru tags, and leaning into niche styles like furry, Guofeng, pets, scenic backgrounds, etc.
flux-3-dev
FLUX 3 Dev is the faster, lower-cost variant of Black Forest Labs' FLUX 3 frontier model, planned for open-weight release. It trades a small amount of peak fidelity for significantly reduced latency and price, making it well suited for rapid iteration, prototyping, and high-volume text-to-image generation.
leonardoai-phoenix-1.0
LeonardoAI Phoenix 1.0 is a professional-grade AI image model designed for realistic, cinematic, and highly detailed visuals. It excels at interpreting complex prompts, rendering text within images, and creating high-resolution outputs suitable for editorial, commercial, or creative projects.
grok-imagine-text-to-image
Grok Imagine is xAI’s high-quality image generation model that transforms text prompts into detailed, stylish, and visually expressive images. It excels at creating vivid scenes, characters, environments, and concept art with strong lighting, depth, and artistic clarity. Get 6 images each time.
nano-banana-pro
Nano Banana 2 is the next-generation image generation developed by Google DeepMind, following the original Nano Banana (also known as Gemini 2.5 Flash Image). It offers advanced text-to-image capabilitie with improved resolution.
qwen3-pro-text-to-image
Qwen 3.0 Pro Text to Image generates high-fidelity professional photorealistic images and editorial artwork from prompts with enhanced detail.
kling-o1-text-to-image
Kling O1 Text-to-Image is a high-fidelity creative image model that converts rich natural-language prompts into ultra-detailed stills. It excels at cinematic composition, realistic lighting, and coherent scene detail—great for concept art, environment renders, character portraits, and stylized imagery with photoreal or illustrative looks.
flux-2-dev
Flux 2 Dev is a powerful text-to-image diffusion model designed for high-quality, fast, and highly detailed visual generation. It excels at creating cinematic lighting, vibrant compositions, surreal concepts, characters, products, and worlds with strong prompt following and artistic control. Ideal for rapid image ideation, visual storytelling, and concept art.
flux-2-pro
Flux-2-Pro Text-to-Image is a premium, high-fidelity generative model capable of producing ultra-realistic, cinematic, and deeply detailed images from text prompts. It excels at complex lighting, layered compositions, surreal visual concepts, and professional art-grade rendering suitable for concept art, advertising visuals, and world-building.
vidu-q2-text-to-image
VIDU Text-to-Image Q2 is a high-quality generative model focused on producing vivid, dynamic, and cinematic still images using natural language prompts. It excels at atmospheric depth, expressive lighting, surreal concepts, and motion-infused compositions typical of VIDU’s visual identity.
reve-text-to-image
Generate images from text prompts using reve's vision capabilities. Ideal for basic concept visuals, diagrams, and abstract compositions.
bytedance-seedream-v4.5
Seedream-v4.5 is ByteDance’s advanced text-to-image diffusion model designed for generating high-detail, high-contrast, cinematic and stylized images. It excels at surreal fantasy concepts, sci-fi worlds, product visuals, photoreal scenes, and artistic compositions with strong prompt adherence and crisp detail.
wan2.6-text-to-image
WAN 2.6 Text-to-Image generates detailed, cinematic still images from text prompts. It focuses on strong composition, atmospheric lighting, and clear subject structure, making it suitable for fantasy and sci-fi environments, surreal concepts, architectural visuals, and dramatic world-building imagery.
flux-2-klein-4b
Flux-2-Klein-4B is a lightweight, fast text-to-image model optimized for clear subject rendering, good prompt adherence, and efficient generation. It works best with simple compositions, everyday scenes, and cute or friendly visuals, making it ideal for UI graphics, demos, thumbnails, mascots, and quick creative iterations.
bytedance-seedream-v5.0
Seedream 5.0 Lite is ByteDance’s next-generation text-to-image model, delivering high-fidelity AI art with advanced visual reasoning and precise typography. Supporting up to 4K resolution and cinematic detail, it excels at complex scene construction, consistent character generation, and real-time knowledge integration for accurate, contextually relevant visuals.
qwen3-text-to-image
Qwen 3.0 Text to Image generates high-fidelity photorealistic images and editorial artwork from prompts with intelligent prompt rewriting.
grok-imagine-image-2
Grok Imagine Image 2.0 is xAI's next-generation image model, built for precise instruction-following, sharp typography, and detailed layout control. It follows instructions closely down to fine detail and plans layout so dense, multi-part visuals hold together, turning a text prompt into a polished, commercial-ready image in one request.
midjourney-v8
Generate 4 photorealistic images per run with Midjourney V8. Improved coherence and detail over V7. Supports text-to-image and reference image guidance.
midjourney-niji
Generate 4 anime and illustration-style images per run with Midjourney Niji. Optimized for character art, manga, and stylized illustrations. Supports reference image guidance.
midjourney-v7
Generate 4 photorealistic images per run with Midjourney V7. Supports text-to-image and reference image guidance via source_image_url.
grok-imagine-text-to-image-quality
Grok Imagine Quality is xAI's high-fidelity text-to-image mode that prioritizes accuracy and detail over speed. It produces sharper, more visually accurate images with stronger lighting, depth, and artistic clarity. Get 6 images each time.
flux-kontext-dev-t2i
Generates an image from a text prompt, with optional reference image for pose or style guidance. Ideal for controlled, consistent image creation using just a description.
hidream-i1-fast
Optimized for speed, this variant generates images in just a few steps. Ideal for previews, real-time applications, and use cases where fast results are more important than fine detail.
hidream-i1-dev
Optimized for speed, this variant generates images in just a few steps. Ideal for previews, real-time applications, and use cases where fast results are more important than fine detail.
常见问题
哪个 AI 图像生成器 API 最好？
取决于速度、画质、文字渲染和艺术风格等需求。Muapi 将多种模型放在同一个接口中，方便你在代码中直接进行 A/B 测试。
如何进行身份验证？
在每次请求的 x-api-key 请求头中发送 Muapi 密钥，可在 muapi.ai/access-keys 创建。
有免费额度吗？
有。新账户会获得免费积分，可以先在 Playground 中体验模型。