MAI-Image-2.6
Microsoft AI
- Released
- July 31, 2026
- Data date
- September 24, 2026
- Access
- Price
- Per image
MAI-Image-2.6 combines text-to-image generation in Microsoft Foundry with precise image-to-image editing. Microsoft documents targeted changes to objects, layouts, and text, motion-blur cleanup, aspect ratios derived from prompt and image inputs, and current Bing results as image context.
Output is PNG, with at least 768 pixels on each side and at most 2,359,296 pixels in total. The published configuration also lists a 32,000-token context window. MAI-Image-2.6 remains in preview, uses proprietary weights, and is tied to Microsoft Foundry. It is not a downloadable local checkpoint.
Specifications and access
| Specification | Value and source |
|---|---|
| Weights | ProprietarySource |
| Access | APISource |
| Context window | 32,000 tokensSource not specified |
| Maximum output | 2,359,296 pixelsSource |
| Lifecycle | Preview in Microsoft FoundrySource |
| Image capabilities | Text-to-image generation and precise image-to-image editingSource |
| Editing controls | Target objects, layouts, and text, and clean up motion blurSource |
| Output format and size | PNG, at least 768 pixels on each side and at most 2,359,296 pixels in totalSource |
| Aspect ratio | Can choose the aspect ratio from the prompt and image inputsSource |
| Web grounding | Can use current Bing Search results as additional image contextSource |
| API authentication | Microsoft Entra ID or an API keySource |
Published benchmarks
Text to Image Arena
Elo ratings summarize pairwise comparisons within this leaderboard. They are not percentages. The scale does not express absolute quality or linear gaps between models.
higher is betterAs of September 22, 2026Scale: 0 Elo - 1,197 Elo
Methodology and context
- Metric
- elo-rating
- Scale
- No upper bound is documented. The chart spans the recorded values and zero.
- Participants
- 30
- Sample
- 8,215
Image Editing Arena
Elo ratings summarize pairwise comparisons within this leaderboard. They are not percentages. The scale does not express absolute quality or linear gaps between models.
higher is betterAs of September 22, 2026Scale: 0 Elo - 1,180 Elo
Methodology and context
- Metric
- elo-rating
- Scale
- No upper bound is documented. The chart spans the recorded values and zero.
- Participants
- 23
- Sample
- 14,258
Recorded measurement details
Text to Image Arena (Show measurement, test conditions, and source)
- Source value
- 1147
- Score
- 1147
- Metric
- elo-rating
- Unit
- Elo
- Direction
- Higher is better
- Source rank
- 5
- Sample
- 8215
- Confidence interval
- 10
- Retrieved at
- 2026-09-22
- USD price per 1,000 tokens
- 38.9
Image Editing Arena (Show measurement, test conditions, and source)
- Source value
- 1133
- Score
- 1133
- Metric
- elo-rating
- Unit
- Elo
- Direction
- Higher is better
- Source rank
- 3
- Sample
- 14258
- Confidence interval
- 9
- Retrieved at
- 2026-09-22
- USD price per 1,000 tokens
- 38.9
Popularity
44.7 / 100
Published prices
- Representative price
- $0.0389 per imageSource
Compare this model
Show all comparisons (35)
Sources and data date
| Type | Evidence and data date |
|---|---|
| Additional source | Microsoft AI (retrieved September 24, 2026) · Microsoft Foundry MAI-Image-2.6 catalog · Editorial description reviewed October 2, 2026 |
| Additional source | Image pricing (retrieved September 22, 2026) |
| Additional source | Lifecycle data · Microsoft Foundry July and August 2026 updates · Editorial description reviewed October 2, 2026 |
| Additional source | Microsoft AI (retrieved September 24, 2026) · Microsoft Foundry MAI image deployment guide · Editorial description reviewed October 2, 2026 |
| Additional source | Artificial Analysis (retrieved September 22, 2026) |
| Additional source | OpenRouter: Popularity index (derived), CC BY 4.0 |
| Additional source | Vercel AI Gateway: Popularity index (derived), CC BY 4.0 |
| Additional source | Creative Commons Attribution 4.0 |