Building a new class of safer, more capable AI systems we call Humanist Superintelligence: AI that is always aligned, controllable, and in service of humanity.
Another No. 1 spot! MAI-Transcribe-2 takes first place for transcription accuracy in David AI's new benchmark. MAI-Transcribe-2 has the lowest error rate in 18 of 21 languages.
[1/4] How well do speech recognition systems handle real conversations in 21 languages?
Today we're releasing DAI-ASR-I18N, a benchmark that evaluates 14 systems on speech recognition and speaker diarization in natural, unscripted conversation
We now have state-of-the-art pre-trained base models in the 3T-parameter weight class, built entirely in-house at MAI.
Post-training results coming soon -- stay tuned!
Lots of details in our technical blog post.
microsoft.ai/news/frontier-p…
MAI-Image-2.6 and MAI-Image-2.6-Flash from @MicrosoftAI are live on AI Gateway.
Both handle text rendering, photorealism, and targeted edits.
𝚊𝚒 𝚒𝚖𝚊𝚐𝚎 \
-𝚖 𝚖𝚒𝚌𝚛𝚘𝚜𝚘𝚏𝚝/𝚖𝚊𝚒-𝚒𝚖𝚊𝚐𝚎-𝟸.𝟼 \
"𝚅𝚎𝚛𝚌𝚎𝚕 𝚂𝚑𝚒𝚙 𝚂𝙵 𝟸𝟶𝟸𝟼"
vercel.com/changelog/mai-ima…
MAI-Code-1.1-Flash is going local with on-device model calls inside Github Copilot.
High-quality agentic coding on your device, with no inference charge for local model calls.
The next era of the PC is taking shape on Windows. ✨
Windows is becoming the home for hybrid intelligence.
🔐 MXC GA
🧠 Frontier coding models running locally
✨ GitHub Copilot HydraFusion
💻 llama.cpp + Windows ML
⚡ Surface Laptop Ultra + Dev Box preorders
🖥️ New RTX Spark PCs
Listen to this support call using @MicrosoftAI's speech models and LiveKit Agents.
• MAI-Transcribe-2-Streaming hears you
• Gemma 4 on LiveKit Inference reasons and calls tools
• MAI-Voice-2.1-Flash answers
What do you think?
STT: livek.it/E4qoZvU
TTS: livek.it/RsCc37l
MAI-Transcribe-2-Streaming, MAI-Voice-2.1, and MAI-Voice-2.1-Flash from @MicrosoftAI are now live on LiveKit. Transcribe-2-Streaming debuts at #1 on the Artificial Analysis accuracy leaderboard → Pair it with MAI-Voice-2.1-Flash to build efficient and expressive voice agents.
Try them out: docs.livekit.io/agents/model…
MAI-Voice-2.1 from @MicrosoftAI is live on OpenRouter.
MAI's highest-fidelity, most expressive text-to-speech model yet:
• Studio-grade, natural speech.
• Same voice can speak in 23 languages while maintaining native accent.
• $22 per 1M characters
Built for audiobooks, podcasts, narration, and brand audio.
openrouter.ai/microsoft/mai-…
More ways to discover, access, and build with MAI models.
MAI models are now available through Vercel, giving developers another way to bring Microsoft AI models into the products and experiences they’re building.
This includes our existing MAI models plus today’s newest releases: MAI-Transcribe-2-Streaming, MAI-Voice-2.1, and MAI-Voice-2.1-Flash.
We partnered with @MicrosoftAI to bring their secure, transparent, and cost-efficient models to AI Gateway. Available today:
• 𝙼𝙰𝙸-𝚅𝚘𝚒𝚌𝚎-𝟸.𝟷: long-form & fast-reply speech
• 𝙼𝙰𝙸-𝚃𝚛𝚊𝚗𝚜𝚌𝚛𝚒𝚋𝚎-𝟸-𝚂𝚝𝚛𝚎𝚊𝚖𝚒𝚗𝚐: transcription
vercel.com/changelog/microso…
Introducing 3 new models: MAI-Transcribe-2-Streaming, MAI-Voice-2.1 and MAI-Voice-2.1-Flash.
Accurate streaming transcription. Natural speech and less waiting between turns.
Build voice agents that keep the conversation moving!
MAI is hill-climbing 61% faster than any other lab in image generation. We've gained +283 Elo on text-to-image in 10 months to optimise the pareto frontier.
No lab has reached the image frontier faster.
Microsoft AI: 864 → 1147 Elo in nine months. That's 377 points a year — 1.6× the next-fastest climb on the board.
More exciting news from @MicrosoftAI: MAI-Image-2.6 is #2 in Text-to-Image Arena with 1,331 pts, and joins the Pareto frontier!
At $38.90 per 1K output images (~$0.039/image), it outperforms Grok Imagine Image 2.0 while also costing less.
MAI-Image-2.6 ranks #2 in:
- Product, Branding & Commercial Design
- 3D Imaging & Modeling
- Cartoon, Anime & Fantasy
- Art
- Text Rendering
It also ranks #3 in Portraits and Photorealistic & Cinematic Imagery.
Try it in Direct Mode, or vote in Battle Mode to help shape the live leaderboard.
MAI-Image-2.6 by @MicrosoftAI has landed in the Text-to-Image Arena at #2 (1336 pts)!
This release is just 45 pts behind GPT Image 2 (Medium) at #1, and ahead of #3 Grok Imagine Image 2.0 (Low) by 20 pts.
MAI-Image-2.6 is a significant improvement from MAI-Image-2.5 overall at #10 with (1256 pts), and across all categories. MAI will also be making it available on Playground with early API access on Foundry next week. See post below for more details.
Congrats to the @MicrosoftAI team on this release!