88Z AI, 3 June — Microsoft has released seven self-developed MAI models trained from scratch, covering reasoning, image, voice, and coding domains, all using unified data standards and evaluation frameworks.
At a glance
- Released seven self-developed MAI models covering reasoning, image, voice, and coding scenarios
- All models are trained from scratch without distillation, using unified data and evaluation standards
The update
Microsoft has officially launched seven new self-developed MAI models, aiming to comprehensively cover core artificial intelligence application scenarios. The lineup includes MAI-Thinking-1 for logical reasoning, and MAI-Image-2.5 along with its efficient version, MAI-Image-2.5-Flash, for image processing. In the realm of voice interaction, the company introduced the MAI-Voice-2 model and its efficient counterpart, MAI-Voice-2-Flash, alongside the MAI-Transcribe-1.5 speech-to-text model. Additionally, the release features the MAI-Code-1-Flash coding model designed for programming tasks. All models are trained from scratch without any distillation techniques and strictly adhere to unified data standards and evaluation frameworks to ensure consistent performance and reliability.
Sources & editorial note
This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
- Microsoft.aiSource published
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.