11 months ago
153
Azure AI Foundry is similar GPS for your AI ambitions: it simplifies navigating the insane gait of innovation successful the AI scenery truthful you spot wherever you are today, measure what’s new, and plan, execute, and negociate your increasing AI portfolio.
I spent astir of the past decennary successful the Midwest (hello, Chicagoland!) but present unrecorded successful the Seattle country and enactment astatine the Microsoft field successful Redmond. Before moving here, I’d visited the Redmond field dozens of times arsenic a Microsoft spouse and customer, but had lone basal airport-campus-hotel-restaurant navigation skills. Relocating a household of 4 meant a batch of exploring to find my mode to market stores (what bash you mean there’s nary Mariano’s successful Seattle?), schools, doctors, dentists, etc. I couldn’t person done it without GPS: Just plug successful the destination and fto it usher us, crook by turn, connected the quickest, safest route, avoiding postulation and roadwork on the way.
The hallmark of bully tech is taking thing analyzable and stressful (ever been mislaid successful an unfamiliar city?) and making it simple. Looking implicit the database of caller AI capabilities we announced successful conscionable the past month, it’s casual to spot wherefore turn-by-turn guidance is needed successful this rapidly expanding AI landscape. Azure AI Foundry is similar GPS for your AI ambitions: it simplifies navigating the insane gait of innovation successful the AI scenery truthful you spot wherever you are today, measure what’s new, and plan, execute, and negociate your increasing AI portfolio portion optimizing costs. All with the assurance that it ever has the latest AI thing.
This wayfinding provides dexterity to assistance your teams enactment nimble arsenic AI presents caller strategical opportunities and efficiencies. Here are conscionable a fewer caller capabilities successful Azure AI Foundry this period that amusement the worth of managing your AI portfolio connected a azygous do-it-all platform.
Models: More is more
As I constitute this determination are 1,917 models successful the Azure AI Foundry exemplary catalog. Choosing one—or three, since galore AI apps present usage aggregate models—is easier than you mightiness deliberation erstwhile facing an abundance of choice. On Azure AI Foundry, your squad tin rapidly measure models connected quality, cost, and throughput, or filter by industry, inference tasks similar translation oregon substance generation, oregon different cardinal features to rapidly get a abbreviated database of models to consider. New tools similar the Evaluation API in Azure OpenAI Service springiness developers the quality to run automated comparisons between OpenAI models, quantify prime improvements, and accelerate the iteration cycle. Together with the Stored Completions API, the Evaluation API provides a programmatic solution for refining and evaluating AI models with accuracy and control.
That’s important erstwhile you see we announced 8 caller Azure OpenAI Service models successful 1 week—the caller GPT-4.1 exemplary series, three precocious audio models, and the o3 and o4-mini reasoning models. You can fine-tune GPT-4.1 and GPT-4.1-mini to accommodate them to your unsocial concern requirements. And past week we announced GPT-image-1, the latest and astir precocious representation procreation model, arsenic good arsenic the latest GPT-4o-realtime-preview and GPT-4o-mini-realtime-preview API updates including WebRTC support, semantic dependable enactment detection and precocious input sound simplification and more.
Check retired however Costar Group uses Azure OpenAI Service models to alteration their customers’ acquisition of uncovering the close home.
That overmuch newness successful 1 week mightiness look similar an anxiousness accelerator—especially if your squad conscionable built an app utilizing an earlier model—but Azure AI Foundry was designed for the progressive, iterative travel of AI development. It’s casual for teams to rapidly measure caller exemplary pros and cons, driblet the exemplary successful the playground for immoderate hands-on testing, find if it’s better, faster, cheaper, etc., and deploy it if desired. Simple. And we mean to alteration OpenAI models successful Azure AI Foundry the aforesaid time arsenic OpenAI announces them. They are backward-compatible and tin beryllium deployed with a azygous click oregon by adding a caller exemplary with a fewer elemental lines of code.
Azure OpenAI accounts for 35 of the 1,917 models successful our catalog, but is not unsocial successful introducing caller models. Also caller successful our exemplary catalog this month:
The first models successful the Meta Llama 4 herd arrived, including Llama 4 Scout, which specializes in condensing oregon analyzing extended information, similar analyzing each documents successful an endeavor SharePoint room to reply a circumstantial query, oregon speechmaking a multi-thousand-page method manual to supply troubleshooting advice. And Llama 4 Maverick, which specializes in representation and substance knowing with enactment for 12 languages for precise image knowing and originative writing.
An unthinkable magnitude of cognition sits successful record cabinets astir the world, invisible to AI. Mistral OCR, a state-of-the-art Optical Character Recognition (OCR) model, tin work and recognize the operation and nuance of hard transcript documents. It tin extract equations, images, tables, and multilingual substance from scanned contracts, handwritten forms, probe papers, regulatory filings, and much astatine scale, making each that cognition accessible for AI to crushed implicit and your squad to larn from and enactment on.
Optimized for the enterprise, Cohere’s Command A and Embed 4 models signifier a powerhouse for precocious apps similar extracting insights from thousands of fiscal filings, net telephone transcripts, and marketplace research.
Deep Seek got the world’s attraction with the merchandise of its R1 exemplary earlier this year. It’s a almighty and capable, low-cost exemplary with immoderate information challenges requiring cautious management. The Microsoft AI squad developed its ain Deep Seek R1 variant, MAI-DS-R1, that outperforms Deep Seek R1 connected responses to blocked topics and reducing harmful contented portion maintaining the aforesaid reasoning capabilities. You tin find MAI-DS-R1 successful Foundry today.
Go to marketplace faster with Azure AI Services
Azure AI Services are businesslike time-to-market accelerators that assistance your developers physique intelligent apps without needing nonstop AI oregon information subject skills. Consider cause building. Until recently, creating agents was analyzable and consumed a batch of developer time. With Azure AI Agent Service successful Azure AI Foundry, what primitively took hundreds of lines of codification tin present beryllium done successful conscionable a fewer lines of code. Azure AI Agent Service is utilized by tens of thousands of organizations to make and deploy agents. Customers like Fujitsu, who built an cause to assistance prioritize income leads, and NTT DATA, which built an cause that speeds connection instauration and surfaces lawsuit insights.
Azure AI Search is different large example. This revolutionary retrieval strategy provides highly applicable hunt results astatine scale. It connects firm cognition into AI applications to powerfulness experiences like UBS’s Legal AI Assistant (LAIA). This adjunct helps their ineligible squad execute probe crossed aggregate languages and 26 cardinal documents to rapidly find related clauses to a lawsuit oregon provisions crossed concern funds.
We precocious announced respective improvements enabling your squad to upgrade to the newest retrieval and retention features without disrupting service. These caller features marque Azure AI Search 1 of the astir cost-effective options connected the market, delivering an enterprise-ready, full-featured retrieval strategy with precocious hunt exertion without sacrificing outgo oregon performance. The effect is empowering your app to present high-quality experiences for each idiosyncratic and enactment with nary compromises.
Coca‑Cola turned its iconic Santa into a globe‑trotting conversationalist successful conscionable 60 days acknowledgment to an AI‑ready instauration connected Azure. Using Azure AI Speech successful Azure AI Foundry, the squad built a real‑time exemplary that fto Santa chat with much than a cardinal radical successful 26 languages, powering an immersive Create Real Magic vacation run crossed 43 markets. It’s a masterclass successful however Azure unlocks lightning‑fast innovation astatine planetary scale.
Get determination with Azure AI Foundry
Organizations built caller musculus astir resiliency and accelerating exertion adoption during the pandemic and that conditioning is paying disconnected arsenic the AI epoch shifts the satellite again, reshaping what is possible. Efficiently navigating this fast-changing scenery is indispensable successful staying competitory and gathering differentiation successful a engaged marketplace. Our intent is for Azure AI Foundry to beryllium your trusted guide, simplifying analyzable processes and empowering your squad to execute more. With the latest advancements and a robust platform, you’re equipped to harness the afloat imaginable of AI, driving innovation and ratio crossed your organization. Stay up of the curve with Azure AI Foundry—your GPS for the aboriginal of AI.
About Jessica
Jessica leads Azure Data, AI, and Digital Applications astatine Microsoft. Find Jessica’s blog posts here and beryllium definite to follow Jessica connected LinkedIn.