MAI Models by Microsoft AI

Microsoft AIProprietary
Microsoft AI releases · 22 in the last 12 monthsChangelog →
23 models2025–2026Up to 1M ctxFrom $0.360/1M input

Last refreshed 2026-11-18. Next refresh: weekly.

Details

ResearcherMicrosoft AI
LicenseProprietary
Commercial useRestricted
Models23
Released2025–2026
Max context1M

Capabilities

Vision9 of 23 models
Multimodal14 of 23 models
Reasoning4 of 23 models
JSON / Tool use5 of 23 models

Links

Website

About

Microsoft AI (MAI) is Microsoft's proprietary model family for Copilot and Azure AI Foundry. The lineup now spans reasoning, coding, image generation/editing, speech synthesis, and transcription models, including MAI-Thinking-1, MAI-Code-1, MAI-Code-1-Flash, MAI-Image-2.5, MAI-Image-2.5-Flash, MAI-Voice-2, and MAI-Transcribe-1.5 alongside the earlier MAI image and speech releases.

Decision facts

Best fit
codingimage generationimage editing
Capability starting point
MAI-Thinking-1 with 256K context and reasoning and JSON / Tool use
Lowest tracked input
MAI-Transcribe-1 · $0.360/1M · Microsoft Foundry
Closest related family
360Zhinao

Current Variants

Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.

16 in view7 retired
MAI Code 2Current

Use when the workload needs coding, 1M context, and JSON / Tool use.

2026-11coding1M contextJSON / Tool use

Use when the workload needs coding, 1M context, and JSON / Tool use.

2026-11coding1M contextJSON / Tool use

Use when the workload needs image generation and multimodal inputs.

2026-11image generationmultimodal inputs

Use when the workload needs image editing, multimodal inputs, and vision.

2026-10image editingmultimodal inputsvision

Use when the workload needs text to speech and audio.

2026-10text to speechaudio

Use when the workload needs text to speech and audio.

2026-10text to speechaudio

Use when the workload needs speech recognition, multimodal inputs, and audio.

2026-10speech recognitionmultimodal inputsaudio

Use when the workload needs image generation, 32K context, and multimodal inputs.

2026-09image generation32K contextmultimodal inputs

Use when the workload needs speech recognition, multimodal inputs, and audio.

2026-09speech recognitionmultimodal inputsaudio

Use when the workload needs reasoning, 256K context, and JSON / Tool use.

2026-08reasoning256K contextJSON / Tool use

Use when the workload needs code, 256K context, and reasoning.

2026-08code256K contextreasoning

Use when the workload needs image generation and multimodal inputs.

2026-08image generationmultimodal inputs
MAI-Code-1Current

Use when the workload needs code.

2026-06code

Use when the workload needs code, 256K context, and reasoning.

2026-06code256K contextreasoning

Use when the workload needs speech recognition, multimodal inputs, and audio.

2026-06speech recognitionmultimodal inputsaudio
MAI-DS-R1Current

Use when the workload needs reasoning, 160K context, and 671B parameters.

2025-04reasoning160K context671B parameters

Release Timeline

8 release groups
2026-11
3 current
MAI Code 2
coding1M contextJSON / Tool use
Current
MAI Code 2 Flash
coding1M contextJSON / Tool use
Current
MAI Image 2.7 Flash
image generationmultimodal inputs
Current
2026-10
4 current
MAI-Image-2.7
image editingmultimodal inputsvision
Current
MAI-Transcribe-2-Streaming
speech recognitionmultimodal inputsaudio
Current
MAI-Voice-2.1
text to speechaudio
Current
MAI-Voice-2.1-Flash
text to speechaudio
Current
2026-09
2 current
MAI-Image-2.6-Flash
image generation32K contextmultimodal inputs
Current
MAI-Transcribe-2
speech recognitionmultimodal inputsaudio
Current
2026-08
3 current
MAI-Code-1.1-Flash
code256K contextreasoning
Current
MAI-Image-2.6
image generationmultimodal inputs
Current
MAI-Thinking-1
reasoning256K contextJSON / Tool use
Current
2026-06
3 current · 3 retired
Current
MAI-Code-1-Flash
code256K contextreasoning
Current
MAI-Image-2.5
image generation32K contextmultimodal inputs
Archived
MAI-Image-2.5-Flash
image generation32K contextmultimodal inputs
Archived
MAI-Transcribe-1.5
speech recognitionmultimodal inputsaudio
Current
MAI-Voice-2
text to speechaudio
Replaced
2026-04
3 retired
MAI-Image-2e
image generation32K contextmultimodal inputs
Archived
MAI-Transcribe-1
speech recognitionmultimodal inputsaudio
Archived
MAI-Voice-1
text to speechmultimodal inputsaudio
Replaced
2026-03
1 retired
MAI-Image-2
image generationmultimodal inputs
Archived
2025-04
1 current
MAI-DS-R1
reasoning160K context671B parameters
Current

Replaced By

Keep for legacy integrations; evaluate MAI-Voice-2.1 before new work.

Keep for legacy integrations; evaluate MAI-Voice-2.1 before new work.

Specifications(23 models)

MAI model specifications comparison
ModelReleasedContextParametersVisionMultimodalReasoningJSON / Tool use
MAI Code 22026-111M—NoNoNoYes
MAI Code 2 Flash2026-111M—NoNoNoYes
MAI Image 2.7 Flash2026-11——YesYesNoNo
MAI-Image-2.72026-10——YesYesNoNo
MAI-Voice-2.12026-10——NoNoNoNo
MAI-Voice-2.1-Flash2026-10——NoNoNoNo
MAI-Transcribe-2-Streaming2026-10——NoYesNoNo
MAI-Image-2.6-Flash2026-0932K—YesYesNoNo
MAI-Transcribe-22026-09——NoYesNoNo
MAI-Thinking-12026-08256K1T total / 35B activeNoNoYesYes
MAI-Code-1.1-Flash2026-08256K138B (5B active)YesYesYesYes
MAI-Image-2.62026-08——YesYesNoNo
MAI-Code-12026-06——NoNoNoNo
MAI-Code-1-Flash2026-06256K—NoNoYesYes
MAI-Transcribe-1.52026-06——NoYesNoNo
MAI-DS-R12025-04160K671BNoNoYesNo

Pricing

MAI model pricing by provider
ModelProviderInput / 1MOutput / 1MType
MAI-Code-1-FlashMicrosoft Foundry$0.75$4.5Serverless
MAI Image 2.7 FlashOpenRouter$1.75$10Serverless
MAI Image 2.7 FlashVercel AI Gateway$1.75$10Serverless
MAI Image 2.7 FlashMicrosoft Foundry$1.75$10Serverless
MAI Image 2.7 Flashfal API$1.75$10Serverless
MAI-Thinking-1Microsoft Foundry$2$8Serverless
MAI-Image-2.6Microsoft Foundry$5$38Serverless
MAI-Image-2.6Vercel AI Gateway$5$38Serverless
MAI-Image-2.7OpenRouter$5$34Serverless
MAI-Image-2.7Vercel AI Gateway$5$34Serverless
MAI-Image-2.7Microsoft Foundry$5$34Serverless
MAI-Image-2.7fal API$5$34Serverless