
MAI-Image 2.5 Text-to-Image API by MICROSOFT
Microsoft's flagship text-to-image generation model, designed to create high-quality, visually rich images from natural language prompts.
MAI-Image 2.5 Text-to-Image को Microsoft द्वारा विकसित किया गया है। Atlas Cloud (Atlas Cloud AI LLC द्वारा संचालित) इस तक पहुँच प्रदान करता है, इसका स्वामी नहीं है। सभी ट्रेडमार्क उनके संबंधित स्वामियों की संपत्ति हैं।
MAI-Image-2.5 Text-to-Image
MAI-Image-2.5 is Microsoft's flagship text-to-image generation model, designed to create high-quality, visually rich images from natural language prompts. It uses a diffusion-based generative approach to progressively refine images, enabling strong alignment between the input text and the generated output. Released on June 2, 2026, it ranks among the top-performing image generation models globally.
Key Capabilities
- Photorealistic image synthesis — Generates realistic imagery with consistent visual structure, accurate lighting, depth, and texture, suitable for concept visualization and professional content creation.
- High-fidelity portraits — Produces expressive, natural-looking portraits with accurate facial structure, lighting, and skin texture.
- Accurate text rendering — Significantly improved rendering of legible text within generated images, including labels, posters, packaging, and signage.
- Visual reasoning — Reasons across objects, scene structure, lighting, scale, and spatial positioning to produce consistent outputs even from ambiguous or complex prompts.
- Product, branding & commercial design — Well suited for product imagery, marketing visuals, brand assets, and commercial creative workflows.
- Creative concept visualization — Translates abstract textual descriptions into visually coherent and imaginative outputs.
Pricing
Pricing is based on two components: the input text tokens in the prompt, and a fixed per-image output fee.
| SKU | Description | Unit Price |
|---|---|---|
sku_input_1m_token | Price per 1M input (prompt) tokens | $5.00 |
sku_output_image | Fixed fee per generated image | $0.05 |
Pricing Formula
cost = countTokens(prompt) / 1,000,000 × $5.00 + $0.05
For most prompts (a few hundred tokens), the token cost is negligible and the effective cost is approximately $0.05 per image.
Examples
| Prompt Length | Token Count | Token Cost | Image Fee | Total |
|---|---|---|---|---|
| Short (e.g., 50 tokens) | 50 | ~$0.000250 | $0.05 | ~$0.0503 |
| Medium (e.g., 500 tokens) | 500 | ~$0.002500 | $0.05 | ~$0.0525 |
| Long (e.g., 2,000 tokens) | 2,000 | ~$0.010000 | $0.05 | ~$0.0600 |
For detailed pricing configuration, see models/microsoft/mai/price/microsoft-mai-image-2.5-text-to-image.json.
Best Use Cases
- Marketing & Advertising — Generate product visuals, campaign imagery, and promotional assets.
- Creative Content — Concept art, illustrations, book covers, and editorial imagery.
- E-commerce — Product visualization and lifestyle photography alternatives.
- Presentations — Custom visuals for slides, reports, and pitch decks.
- Prototyping — Rapid visual mockups for design and UX workflows.
- Signage & Packaging — Designs that require legible in-image text rendering.
Pro Tips
- Be specific about lighting, perspective, and style (e.g., "soft golden-hour lighting", "top-down view", "photorealistic").
- Mention the subject first, then environment, then style and mood.
- For text in images, keep inscriptions short and clearly stated in the prompt.
- Use aspect ratios suited to your use case (portrait for people, landscape for scenes).
Technical Specifications
| Spec | Value |
|---|---|
| Model Developer | Microsoft AI |
| Release Date | June 2, 2026 |
| Input Format | Text prompt (natural language) |
| Output Format | PNG image (base64-encoded) |
| Max Prompt Length | 32,000 tokens |
| Max Output Resolution | 1,048,576 total pixels (e.g., 1024×1024) |
| Supported Languages | English (primary) |
Related Models
- MAI-Image-2.5 Edit — Same model with image-to-image editing capability.
- MAI-Image-2.5-Flash Text-to-Image — Faster, lower-cost text-to-image variant.
- MAI-Image-2.5-Flash Edit — Faster, lower-cost image editing variant.



















