Runware, a developer infrastructure platform for real-time generative media, has closed a $50 million Series A led by Dawn Capital, with participation from Comcast Ventures, Speedinvest, Insight Partners, and a16z Speedrun.
The raise brings total funding to $66 million.
From Frustration to Founding
Founder Flaviu Radulescu launched Runware in 2023 after testing early text-to-image tools and noticing a recurring issue: impressive AI models, but slow, inconsistent output. Partnering with Ioana Hreninciuc, the duo built Runware as a high-performance generation layer that delivers images, video, and audio in real time.
Since launching, the platform has powered over 10 billion creations for more than 200,000 developers.
Advertisement
One Unified API for Every Media Model
Runware gives developers a single interface to generate AI media without setting up infrastructure or maintaining dozens of separate model integrations.
The company provides:
- Custom inference infrastructure optimized for open-source models
- Day-zero access to new models as soon as they are released
- Unified API for all media types
- Highly competitive pricing
Hreninciuc, who oversees operations and GTM, says Runware’s advantage comes from its Sonic Inference Engine, built on custom AI hardware and tightly optimized loading/offloading software that supports over 400,000 models for real-time inference.
The company also plugs into cloud AI providers to automatically reroute workloads when higher memory or throughput is required.
A Competitive Market Heating Up
Developer-facing generative media tools have become a hotspot for VC investment. Competitors like Fal.ai (recently raised $140M at a $4.5B valuation) and Replicate target breadth and easy deployment of open-source models.
The startup differentiates with speed, cost control, and a pay-per-output model.
Instead of purchasing GPU time, customers pay for what they generate—similar to Stable Diffusion or Flux—offering more predictable economics for scale.
Building Toward “The API for All AI”
The new capital will go toward expanding Runware’s infrastructure, with a goal of supporting over 2 million models through the Sonic Inference Engine. The team, currently around 25 people, will grow significantly across engineering and research.
“We’re expanding aggressively into new modalities so applications can scale to millions of users without margin pressure,” Hreninciuc said. “Our goal is simple: make powerful generative AI accessible and affordable for everyone—from builders to end users.”
About Runware
Runware provides real-time generative infrastructure for developers, offering a unified API that powers image, video, and audio output at scale. Built on custom hardware and advanced inference optimization, Runware enables developers to integrate and run hundreds of thousands of open-source models with high performance and low cost.
