Microsoft has made its Fireworks AI integration generally available on Azure Foundry, giving startups access to 26 open-weight models and allowing Microsoft for Startups members to use eligible Azure credits for deployments. The program advertises up to $150,000 in credits.
The catalog includes models from DeepSeek, Moonshot AI, Z.ai, MiniMax, Qwen, Google and OpenAI’s open-weight gpt-oss line. Fireworks handles inference, while Foundry provides model governance, discovery and billing.
Microsoft’s deployment blueprint, published on August 4, 2026, is designed to run within a startup’s own Azure subscription. It connects an application on Azure Container Apps to a Fireworks model endpoint, with Azure Container Registry for images and Key Vault for credentials. Startups can later add API Management, Redis caching and Azure Monitor as usage grows.
The company said the architecture is intended to help teams begin with a serverless model endpoint rather than manage their own GPU infrastructure. However, the credit benefit applies only to pay-per-token Data Zone Standard usage and excludes provisioned throughput units.
Serverless deployments are currently limited to six US Azure regions. Microsoft also notes that the service falls outside its EU Data Boundary commitments, has no FedRAMP authorization and cannot process payment-card data. Customers remain responsible for assessing the safety and behavior of the models.
Source: Unite.AI
