Free tools Windows power users keep installed
One-click scans. No signup required.
ThatPainter is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Learn More
MAI-Image-1 is Microsoft’s first image-generation model that the company describes as developed entirely in-house. Microsoft announced it on October 13, 2025, initially introduced it for public testing through LMArena, and began integrating it into Bing Image Creator and selected Copilot experiences on November 4, 2025.
It is important to distinguish the model from the products that expose it: MAI-Image-1 is not a standalone consumer app. Readers typically encounter it through Bing Image Creator, while Microsoft’s current documentation also references newer MAI image variants. As of August 18, 2026, MAI-Image-1 is best understood as a significant launch milestone—not necessarily Microsoft’s newest image model.
What is MAI-Image-1?
MAI-Image-1 is a text-to-image model in Microsoft’s MAI family. Microsoft presented it as the third purpose-built model in that family, following MAI-Voice-1 and MAI-1-preview, and as its first image-generation model developed entirely in-house.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →The model is integrated into Microsoft services rather than distributed as a separate image-generation application. At launch, Microsoft said it was available alongside models including DALL-E 3 and GPT-4o in Bing Image Creator. That positioning signaled diversification: Microsoft was building more of its own AI stack while continuing to use partner models.
MAI-Image-1’s launch timeline
- October 13, 2025: Microsoft announced MAI-Image-1 and said it had debuted among the top 10 text-to-image models on LMArena. The model was initially available for public testing there.
- November 4, 2025: Microsoft announced integration into selected products, including Bing Image Creator and Copilot Audio Expressions’ Story Mode.
- August 18, 2026: Microsoft’s current Bing Image Creator documentation references newer MAI variants, including MAI-Image-2e. The model menu may therefore differ from the launch-era lineup.
Microsoft said availability extended to countries with access to Bing Image Creator and Copilot Labs. That does not mean universal availability: the interface, account requirements, model choices and rollout can vary by region, device, account type and date.
What Microsoft says it can do well
Microsoft’s launch materials emphasize the following intended strengths:
- Photorealistic imagery.
- Natural lighting, bounce light and reflections.
- Landscapes and other natural scenes.
- Food imagery.
- Visual diversity.
- Relatively fast generation for rapid iteration.
The company’s example gallery focuses heavily on nature and food, including water, animals, mountains, fields, sunsets, pizza, sushi and fruit. Those examples illustrate Microsoft’s priorities, but they are not an independent benchmark. The careful conclusion is that Microsoft positioned MAI-Image-1 for attractive, natural-looking image generation—not that it has been proven superior to Midjourney, Imagen, GPT Image, DALL-E or every other current model.
Microsoft also said it used careful data selection and nuanced evaluation to reduce repetitive or generically stylized results. Creative-industry professionals provided feedback during development and evaluation. Those statements do not establish that artists approved the training data, that the model avoids all stylistic imitation or that it is free from copyright disputes.
What the LMArena ranking means
Microsoft’s claim that MAI-Image-1 debuted in LMArena’s top 10 is a useful early signal, but it needs context. LMArena’s text-to-image leaderboard is based on user preferences in anonymous pairwise comparisons. A strong debut suggests that participants preferred its outputs in those comparisons at that time.
It is not the same as a controlled technical benchmark. Rankings change as votes accumulate and new models are added. The result does not by itself establish performance for:
- Legible text, logos or packaging.
- Hands, anatomy and complex interactions.
- Character consistency across a series.
- Exact product reproduction.
- Editing and image-to-image workflows.
- Print resolution or commercial production.
- Copyright clearance or commercial licensing.
In short: Microsoft’s claim was “top 10 at launch”; the defensible interpretation is “promising early reception”; the unsupported conclusion is “the best image generator.”
How to try MAI-Image-1 in Bing Image Creator
- Open bing.com/create.
- Sign in with a personal Microsoft Account.
- Start a new text-to-image prompt.
- Open the model selector, if it is available in your interface.
- Select MAI-Image-1 or the currently listed MAI variant.
- Submit the prompt and choose from the generated results.
Microsoft’s Bing documentation describes Image Creator as free for users with a Microsoft Account. It currently documents 15 free fast creations per day, a slower standard mode and up to 200 prompts in a 24-hour period. Microsoft Rewards points may provide additional fast generations after the daily allowance, subject to current account and program rules.
These are product limits documented by Microsoft, not a permanent MAI-Image-1 pricing plan. Limits and model availability may change. Microsoft does not present MAI-Image-1 in the reviewed sources as a standalone paid subscription, public API or downloadable model.
Account and access restrictions
Microsoft says Bing Image Creator requires a Microsoft Account and is not available to users signed in with a Microsoft Entra ID account, which commonly means a work or school account. If the model selector is missing, possible explanations include regional availability, account type, a different Microsoft product surface, a model rollout change or replacement by a newer MAI model.
Generation is not the same as editing
A particularly important practical distinction concerns uploaded images. Microsoft’s Bing Image Creator documentation says that editing or transforming an uploaded image is handled by GPT-4o, not MAI-Image-1 or DALL-E 3.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteTherefore, choosing an MAI model for a new image from text does not mean that every later editing operation uses that model. The exact workflow and selector can differ in the current interface, but users should not assume that MAI-Image-1 powers all Bing Image Creator features.
How it compares with Microsoft’s other image choices
| Option | Best described as | Key distinction |
|---|---|---|
| MAI-Image-1 | Microsoft’s first fully in-house image model | A launch milestone positioned around photorealism, lighting and speed |
| DALL-E 3 | OpenAI image model available in Bing Image Creator at launch | An external model offered alongside Microsoft’s model |
| GPT-4o image generation and editing | OpenAI model used in selected Microsoft workflows | Microsoft says uploaded-image editing uses GPT-4o |
| Newer MAI variants | Later Microsoft image models | They may replace or supersede MAI-Image-1 in current interfaces |
This lineup does not demonstrate that Microsoft has stopped relying on OpenAI. It demonstrates that Microsoft is combining internally developed models with partner models and exposing some of them through the same consumer workflow.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Why the in-house claim matters
Developing its own image model can give Microsoft more control over product integration, model behavior, customization, inference costs, speed, safety policies and deployment decisions. It also reduces the need to treat a single external provider as the only source of image-generation capability.
“Developed entirely in-house” should still be read precisely. Microsoft’s wording describes the model’s development; it does not prove that every part of its training data, infrastructure or research process was created without outside technology. The claim also does not mean Microsoft’s products no longer use OpenAI models.
What remains unclear
Microsoft’s announcement offered high-level safety and responsible-development statements and said public testing would help gather feedback before broader deployment. The sources reviewed do not provide a complete MAI-Image-1 model card, detailed training-data disclosure, quantitative safety evaluation or comprehensive list of prohibited prompts.
There is also no basis in the supplied evidence for calling MAI-Image-1 commercially safe, copyright-safe or reliably suitable for professional production. Anyone using generated images for advertising, products, publishing or client work should check Microsoft’s current terms and independently assess rights, originality, consistency and output quality.
Common failure modes
- The model option is missing: Check whether the interface has changed, whether the account and region are supported, and whether Microsoft has replaced the model with a newer MAI variant.
- A work or school account fails: Try a personal Microsoft Account; Microsoft says Entra accounts are not supported for Bing Image Creator.
- An uploaded-image edit uses GPT-4o: That is consistent with Microsoft’s current documentation rather than evidence of a malfunction.
- Fast generations are exhausted: Wait for standard generation or use available Rewards options, subject to current limits.
- Results look generic or inconsistent: Microsoft’s design goals are not guarantees. More specific prompts and simpler compositions may help, but no model is equally reliable across all subjects.
- One prompt class performs better than another: The launch examples favor landscapes, nature and food. Do not infer equivalent performance for typography, logos, hands, dense scenes or multi-image consistency.
Bottom line
MAI-Image-1 matters primarily because it showed Microsoft building and deploying a specialized image model of its own inside high-traffic products. Its top-10 LMArena debut was an encouraging launch signal, and Microsoft’s examples support its focus on photorealistic scenes, lighting, landscapes and food.
For readers, Bing Image Creator remains the practical place to try Microsoft’s image models for free with a personal account. But the model menu is not static, and Microsoft’s current documentation references newer MAI variants. Treat MAI-Image-1 as Microsoft’s first in-house image-generation milestone—not automatically its latest, best or only image model.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
