Last checked: July 10, 2026
MiniMax Image 01 is an AI image generation model from MiniMax, built for turning text prompts into images and, through the official API, generating new images from a single subject reference image. It is most useful for developers, marketers, creators, and product teams who want low-cost image generation with API access, batch output, multiple aspect ratios, and optional reference-image workflows.
MiniMax introduced Image-01 in a February 28, 2025 announcement as its first text-to-image generation model, while MiniMax’s release notes list Image-01 under February 15, 2025. Its official documentation now describes both Text-to-Image and Image-to-Image capabilities for the image generation service.
Quick facts about MiniMax Image 01
| Item | Current details |
|---|---|
| Model name | image-01 |
| Creator | MiniMax |
| Main use | Text-to-image generation and image generation with a reference image |
| Official endpoint | POST /v1/image_generation |
| Prompt limit | 1,500 characters in the official MiniMax API |
| Images per request | 1–9 images |
| Official pay-as-you-go price | $0.0035 per image, according to MiniMax’s current pricing page |
| Official image API rate limit | 10 requests per minute for Image Generation image-01 |
| Output formats | URL or base64; MiniMax notes that URL outputs expire in 24 hours |
| Best fit | Affordable API image generation, marketing visuals, character/reference workflows, creative variation, product concepts |
| Check carefully | Commercial-use terms, copyright risk, third-party provider pricing, and current API limits |
MiniMax’s official text-to-image API documentation lists image-01 as the model option, supports a 1,500-character prompt, offers multiple aspect ratios, allows custom width and height from 512 to 2048 pixels when used together, and supports url or base64 response formats. The same docs state that n can generate 1–9 images per request and that prompt_optimizer defaults to false. MiniMax’s pricing page lists image-01 at $0.0035 per image, while the rate-limit page lists Image Generation image-01 at 10 RPM.
What is MiniMax Image 01?
MiniMax Image 01, also written as MiniMax Image-01 or image-01, is MiniMax’s image generation model for creating images from text prompts. MiniMax’s launch announcement describes it as the company’s first text-to-image model and highlights prompt control, visual composition, realistic human and object rendering, and support for standard aspect ratios such as 16:9, 4:3, 3:2, 2:3, 3:4, 9:16, and 21:9.
The easiest way to understand it is this: MiniMax Image 01 is not a general chatbot model and not the same thing as MiniMax’s Hailuo video models. It is the image generation model exposed through MiniMax’s image API and through third-party providers. Hailuo is MiniMax’s video product family, while Image 01 is the image model used for still-image generation. MiniMax’s main site lists Hailuo under its video products and presents MiniMax as a company developing multimodal models across text, audio, image, video, and music.
What can MiniMax Image 01 do?
MiniMax Image 01 can generate images from written prompts. In practical terms, that means you can describe a scene, product, person, environment, art direction, lighting style, camera angle, color palette, or format, and the model returns one or more images. The official API guide describes Text-to-Image as creating images directly from detailed text descriptions.
It can also generate images using a reference image. MiniMax’s guide says the reference-image feature lets users provide a clear subject image, including an online image URL, then combine it with a text prompt to generate a new image that preserves key characteristics of the subject. The guide also states that only a single reference image is supported per request.
That makes MiniMax Image 01 especially relevant for:
- Marketing visuals: social posts, banners, product concepts, landing-page hero images, and ad mockups.
- Character workflows: generating variations of the same virtual character from a reference image.
- Product visualization: placing products in new settings, styles, or campaign concepts.
- Creative direction: generating multiple visual routes before commissioning final assets.
- Developer products: adding image generation into apps, automation tools, creative platforms, or internal workflows.
It should not be treated as a guaranteed replacement for professional photography, legal clearance, brand design, or final production art review. AI image outputs can vary, and reference-image similarity does not guarantee perfect identity preservation.
MiniMax Image 01 API: how it works
The official MiniMax API uses the image generation endpoint with model: "image-01". The documentation shows examples for both text-to-image generation and subject-reference image generation, in which one reference image is combined with a text prompt to preserve key characteristics of the subject.
Core API parameters
| Parameter | What it controls | Official details to know |
|---|---|---|
model | The image model | Use image-01 for MiniMax Image 01 |
prompt | The text description | Required; max length 1,500 characters |
aspect_ratio | Output format | Defaults to 1:1; supports 1:1, 16:9, 4:3, 3:2, 2:3, 3:4, 9:16, 21:9 |
width / height | Custom dimensions | 512–2048 px, divisible by 8; must be set together; aspect_ratio takes priority if both are provided |
response_format | Returned image format | Defaults to url; options are url and base64; URLs expire in 24 hours |
seed | Reproducibility | Controls generation randomness and can improve repeatability when the same seed and parameters are reused, but identical output should not be assumed. |
n | Number of images | 1–9 images per request; default is 1 |
prompt_optimizer | Prompt enhancement | Boolean; default is false |
subject_reference | Reference image workflow | Used for image-to-image/reference generation |
These parameter details come from MiniMax’s official text-to-image and image-to-image documentation.
Simple Python API example
import os
import requests
url = "https://api.minimax.io/v1/image_generation"
headers = {
"Authorization": f"Bearer {os.environ['MINIMAX_API_KEY']}",
"Content-Type": "application/json",
}
payload = {
"model": "image-01",
"prompt": (
"A premium skincare bottle on a warm stone bathroom counter, "
"soft morning light, realistic product photography, clean background, "
"natural shadows, 16:9 composition"
),
"aspect_ratio": "16:9",
"response_format": "url",
"n": 3,
"prompt_optimizer": True,
}
response = requests.post(url, headers=headers, json=payload, timeout=60)
response.raise_for_status()
print(response.json())Use environment variables for API keys and avoid exposing keys in frontend code, public repositories, or browser requests. MiniMax’s API FAQ warns users not to share API keys or expose them in browser/client-side code.
MiniMax Image 01 pricing
MiniMax’s official pay-as-you-go pricing page currently lists image-01 at $0.0035 per image. That is the official MiniMax API price checked for this article, not a permanent guarantee and not necessarily the price charged by third-party providers.
Third-party providers may charge differently because they add their own hosting, queueing, SDK, infrastructure, margin, and billing model. Fal’s MiniMax Image 01 page says one image costs $0.01, while Replicate’s public listing says the model costs $0.01 per output image. Segmind provides a serverless API page for image-01 with parameters such as prompt, aspect_ratio, number_of_images, and prompt_optimizer, while WaveSpeed provides separate MiniMax Image 01 text-to-image and reference-image listings, each with its own API wrapper and live pricing display.
| Access route | What to expect | Pricing note |
|---|---|---|
| Official MiniMax API | Direct access to MiniMax’s Open Platform image API | MiniMax lists image-01 at $0.0035 per image |
| Fal | Hosted API/playground wrapper for MiniMax Image 01 | Fal’s page shows $0.01 for one image |
| Replicate | Hosted model page with API and playground | Replicate’s listing says $0.01 per output image |
| Segmind | Serverless API wrapper | Check live Segmind pricing before production use |
| WaveSpeed | Hosted text-to-image and image-to-image pages | Check the live cost preview because provider pricing may vary by parameters |
For production budgeting, calculate cost using your real usage pattern: number of generated images, retry rate, failed outputs, batch size, provider fees, storage, moderation, and human review. A cheap per-image price can become expensive if your workflow generates many unused variations.
Best use cases for MiniMax Image 01
1. Fast campaign concepting
MiniMax Image 01 is a good fit when a marketing team needs several visual directions before committing to a shoot, designer, or final art direction. Use it to explore lighting, settings, backgrounds, seasonal themes, and social formats.
Example prompt:
A clean editorial product campaign for a reusable stainless steel water bottle on a desert hiking trail, golden-hour light, realistic photography, dust texture, premium outdoor brand style, no text, 16:9.
2. Character variation with a reference image
The reference-image feature can help preserve recognizable subject characteristics across multiple generated scenes, although consistent identity is not guaranteed and should be tested across repeated outputs. MiniMax’s guide specifically frames reference images as useful for consistent visual identity, such as generating the same virtual character in different contexts.
Example prompt:
Using the provided character reference, generate the same character sitting in a cozy bookstore cafe, warm window light, casual winter outfit, natural expression, realistic editorial photography, shallow depth of field.
3. Product and ecommerce mockups
Image 01 can be used to mock up product environments, lifestyle settings, packaging directions, and advertising concepts. Keep final ecommerce assets subject to brand, legal, and quality review.
Example prompt:
A minimal ceramic coffee mug on a modern kitchen counter, soft natural daylight, steam rising, Scandinavian interior, realistic product photography, clean composition, no logos, square crop.
4. Developer image generation features
For developers, MiniMax Image 01 is most relevant when integrated into tools that need controlled output formats, partially repeatable generation through seed values, and low-cost batch output. The official API supports seed, n, response format control, and multiple aspect ratios, which are useful for repeatable workflows and productized image generation.
5. Creative thumbnails and social assets
MiniMax Image 01 can generate multiple sizes for platforms such as YouTube thumbnails, TikTok covers, Instagram posts, blog headers, and ad variants. Use aspect ratios intentionally: 16:9 for thumbnails and hero banners, 9:16 for vertical video covers, and 1:1 for square social assets.
Prompt tips for better MiniMax Image 01 results
MiniMax’s official launch announcement emphasizes prompt-to-image fidelity, visual composition, lighting, environments, and realistic subject/object rendering. In practice, prompts are usually stronger when they specify the subject, setting, composition, lighting, style, output purpose, and exclusions.
Use this structure:
[Main subject] + [context/setting] + [composition] + [lighting] + [style/medium] + [technical details] + [constraints]Better prompt pattern
Instead of:
Make a nice product photo.
Use:
A premium glass perfume bottle on black marble, close-up product photography, dramatic rim lighting, soft reflection, dark luxury background, shallow depth of field, no text, no logo, 4:3.
Prompt checklist
Before generating, define:
- The main subject.
- The visual purpose.
- The aspect ratio.
- The realism or art style.
- Lighting and mood.
- Camera angle or composition.
- What should not appear.
- Whether text is required.
- Whether a reference image is needed.
Avoid overloading the prompt with contradictory styles such as “minimalist, maximalist, cyberpunk, natural daylight, dark studio, watercolor, photorealistic” in the same request.
Limitations and risks
MiniMax Image 01 is useful, but it is not the best choice for every image workflow.
Text inside images may need another model
MiniMax’s official Image 01 documentation does not position the model primarily as a typography or text-rendering specialist. If the asset depends on accurate words inside the image, such as signs, packaging, posters, labels, or logo text, test carefully. Ideogram markets text rendering as a core capability and claims 95% text accuracy on its text-rendering feature page, while OpenAI’s current image-generation documentation lists GPT Image models, including gpt-image-2, for image generation and editing through the API. Model availability, pricing, and access requirements should be checked before production use.
Reference image support is not unlimited
MiniMax’s official guide says only one reference image is supported per request. That is useful for a single subject, but it may be limiting for workflows that need multiple product references, multi-character consistency, or precise scene reconstruction.
API throughput may matter
The official MiniMax rate-limit page lists Image Generation image-01 at 10 requests per minute. That may be enough for individual creators and small apps, but production systems should queue jobs, handle retries, and contact MiniMax or a provider if higher throughput is required.
Commercial use and copyright need careful review
Do not assume that every generated image is automatically cleared or risk-free for commercial use. MiniMax’s general Terms of Service state that specific services may be governed by additional product terms, and that product-specific terms control in the event of a conflict. API users should review the current MiniMax Open Platform terms and privacy policy for their exact access route before uploading confidential brand assets, private likenesses, customer images, or unreleased product photos.
Copyright risk is especially important for generative image tools. Reuters reported in September 2025 that Disney, Universal, and Warner Bros. Discovery sued MiniMax over alleged copyright infringement involving Hailuo AI, and Reuters later reported in May 2026 that a U.S. federal judge denied MiniMax’s bid to dismiss the lawsuit. These are allegations and procedural developments, not final legal conclusions, but they are a strong reason to avoid prompts involving copyrighted characters, studio franchises, logos, celebrities, or brand assets unless you have the rights to use them.
MiniMax Image 01 vs alternatives
| Tool/model | Better fit when you need… | Watch out for… |
|---|---|---|
| MiniMax Image 01 | Low-cost API image generation, reference-image workflows, multiple aspect ratios, batch outputs | Commercial terms, reference-image limits, text rendering uncertainty, rate limits |
| Midjourney | Highly stylized creative art direction and polished visual exploration | Subscription workflow and plan limits; verify current model/version details |
| Ideogram | Typography-heavy visuals, posters, logos, packaging, readable text | Pricing and licensing depend on plan/API terms |
| OpenAI GPT Image models | Image generation and editing inside OpenAI API workflows | Token/image pricing and model availability can change |
| FLUX / Black Forest Labs | Developer API access, FLUX model ecosystem, production image generation workflows | Model choice, deployment, and pricing vary by route |
| Adobe Firefly | Commercially oriented creative workflows inside Adobe’s ecosystem | Features, indemnification, and model availability depend on plan and terms |
Midjourney’s documentation currently lists V8.1 as the default version as of June 10, 2026. Because Midjourney updates its default models over time, users should verify the active version, available features, and plan limits before making direct comparisons with MiniMax Image 01. Ideogram is worth evaluating when accurate text rendering is a priority, as its official documentation highlights typography-focused workflows such as posters, packaging, signage, book covers, and logos. OpenAI is a practical alternative for teams already building on the OpenAI API, with its current image-generation documentation covering GPT Image models, including gpt-image-2, for image generation and editing. Black Forest Labs provides official FLUX API documentation for text-to-image generation, while Adobe positions Firefly for commercially oriented creative workflows and offers IP indemnification for certain eligible plans and use cases, subject to the applicable terms.
Who should use MiniMax Image 01?
MiniMax Image 01 is worth considering if you need:
- A low-cost official image API.
- Text-to-image generation with common aspect ratios.
- Up to 9 images per request.
- A reference-image workflow for a single clear subject.
- URL or base64 output.
- A practical model for marketing, product concepts, creative testing, or app integration.
Consider another model if you need:
- Best-in-class typography inside images.
- Multiple reference images in one request.
- A mature design-suite workflow.
- Built-in enterprise governance and indemnification.
- Local deployment or open-weight customization.
- Very high throughput without rate-limit planning.
- Final commercial assets involving protected characters, trademarks, or likenesses.
Practical workflow: from prompt to publishable asset
A reliable MiniMax Image 01 workflow should look like this:
- Define the job: ad concept, product mockup, character image, social post, or app output.
- Choose the format:
1:1,16:9,9:16, or custom dimensions if needed. - Write a precise prompt: include subject, setting, composition, lighting, style, and exclusions.
- Generate variations: use
nto create multiple options. - Review quality: check hands, faces, product details, unwanted logos, distorted objects, and artifacts.
- Check legal risk: avoid protected characters, real people without permission, trademarks, and confidential assets.
- Edit manually if needed: use a design tool for typography, layout, retouching, and brand compliance.
- Save outputs properly: if using URL responses, download or store them before the 24-hour URL expiry noted in the MiniMax docs.
FAQ
What is MiniMax Image 01?
MiniMax Image 01 is MiniMax’s AI image generation model for creating images from text prompts. The official MiniMax API documentation also supports reference-image generation through the image generation service.
Is MiniMax Image 01 free?
The official MiniMax pay-as-you-go pricing page lists image-01 at $0.0035 per image, so API use should be treated as paid unless MiniMax or a third-party provider offers credits or a temporary promotion.
Does MiniMax Image 01 have an API?
Yes. MiniMax’s official documentation shows POST /v1/image_generation with model: "image-01" for text-to-image and image-to-image/reference workflows.
Does MiniMax Image 01 support reference images?
Yes. MiniMax’s guide says users can supply a reference image with a clear subject and combine it with a text prompt to generate a new image that preserves key characteristics. The same guide says only one reference image is supported per request.
How many images can MiniMax Image 01 generate per request?
The official API documentation lists n as the number of images to generate per request, with a range of 1 to 9 and a default of 1.
What aspect ratios does MiniMax Image 01 support?
The official docs list 1:1, 16:9, 4:3, 3:2, 2:3, 3:4, 9:16, and 21:9. They also list corresponding default dimensions for these aspect ratios.
Can I use MiniMax Image 01 commercially?
Do not assume commercial rights without checking the current terms for the exact access route you use. MiniMax’s general Terms of Service say product-specific terms apply to specific services, and third-party providers may add their own terms.
Is MiniMax Image 01 better than Midjourney, Ideogram, or GPT Image?
Not universally. MiniMax Image 01 is attractive for low-cost API image generation and reference-image workflows. Ideogram is worth evaluating for typography-heavy assets, GPT Image models may suit teams already using OpenAI’s image-generation and editing APIs, and Midjourney remains an option for highly stylized creative exploration. Results should be compared using the same prompts, formats, and review criteria.