Microsoft launches Fireworks AI blueprint for startups on Azure Foundry

Microsoft has made its Fireworks AI integration generally available on the Azure Foundry platform, offering a reference architecture that lets startups run open‑weight AI models with high throughput and low latency. The blueprint walks teams through deploying a model endpoint, routing traffic via Azure API Management, adding Redis caching, and monitoring latency, errors and token usage with Azure Monitor, all while keeping governance and billing in a single Azure control plane.

The catalog now lists 26 open models from providers such as DeepSeek, Moonshot AI, Z.ai, MiniMax, Qwen, Google and OpenAI’s open‑weight GPT‑OSS line. Six models are offered with serverless pay‑per‑token pricing, while the rest use reserved‑capacity units. Startups in the Microsoft for Startups program can apply up to $150,000 in Azure credits toward these deployments, eliminating the need to manage their own GPU infrastructure.