Oct 16, 2025

Runpod Launches Public Endpoints for AI Model Access By Spencer Hulse Spencer Hulse has been verified by Muck Rack's editorial team Published on October 15, 2025 Runpod is a cloud platform specializing in GPU-based infrastructure for artificial intelligence applications. The company provides on-demand access to graphics processing units across dozens of global data center regions. Its services have attracted a growing user base by focusing on the needs of AI developers. By mid-2024, Runpod’s platform had over 100,000 developers using its GPU instances and serverless endpoints. (The user base has since tripled; Runpod reports more than 500,000 developers on the platform by mid-2025.) The startup is backed by major tech investors, including Intel Capital and Dell Technologies Capital, which co-led a $20 million seed funding round in 2024. Runpod’s main offerings prior to 2025 were GPU Cloud (allowing users to spin up GPU-powered virtual machines for training or other compute tasks) and Serverless deployment, which lets developers create scalable API endpoints for their AI models without managing servers. These features are designed to simplify AI development: a developer can launch a container with a chosen model and get a live API for inference in production within minutes. This focus on developer experience, making infrastructure setup so straightforward that users can “set and go,” has been central to Runpod’s strategy. According to co-founder and CEO Zhen Lu, enabling rapid iteration for developers is what “matters most” in unlocking value from AI projects. Public Endpoints Launch in 2025 On August 6, 2025, Runpod officially announced the launch of a new feature called Public Endpoints, expanding its platform to offer ready-to-use AI models accessible via API. Public Endpoints provide instant access to advanced AI models through simple HTTP requests, with an API playground available for testing through the Runpod Hub. This means that instead of requiring users to set up their own model servers, Runpod hosts a selection of popular models that developers can invoke on demand. Public Endpoints span multiple AI domains. ByteDance, the company behind TikTok, partnered with Runpod to showcase its generative video model Seedance 1.0 Pro running on the platform at the launch event. Another model from ByteDance, Seedream 3.0, provides advanced image generation and is also included in the endpoint lineup. The service features large language models and speech recognition tools as well: a 70-billion-parameter Llama 2 variant (dubbed Deep Cogito v2) is offered, and OpenAI’s Whisper speech-to-text model is among the available endpoints. All these endpoints are billed on a usage-based model, allowing developers to tap into powerful AI models without provisioning their own servers. Runpod even offered free credits during the launch period to encourage users to experiment with every available endpoint. Competition and Industry Context Runpod’s rollout of Public Endpoints comes amid a broader trend of specialized AI infrastructure providers emerging to meet the exploding demand for machine learning and generative AI workloads. Industry observers have noted that general-purpose cloud platforms often struggle with the latency and scaling needs of AI, which has led to a new wave of cloud services built specifically for AI tasks. Runpod is one of several startups in this space. Its competitors have been attracting significant investment: CoreWeave, a New Jersey-based GPU cloud provider, went public in March 2025 at $40/share and has since grown to ~$130/share. Another rival, Lambda Labs, raised $480 million at a $4  billion valuation to expand its AI-focused cloud platform. This influx of capital underscores the surging demand for GPU-as-a-service. One market analysis forecasts that the GPU cloud market will grow from about $3.3 billion in 2023 to $33.9 billion by 2032. In this competitive landscape, Runpod’s strategy has been to iterate quickly on features developers ask for, such as easier model deployment workflows, in order to stand out. The introduction of Public Endpoints in 2025 aligns with that approach, offering a library of ready-made AI models alongside the option for users to deploy their own. The move also shows how independent cloud platforms can partner with AI research labs (in this case, ByteDance’s) to bring advanced AI models to a wider user base. For developers and businesses, services like Runpod’s Public Endpoints are making advanced machine learning models more accessible as cloud APIs. Tags