Baseten Deploys Runway Video Models to Dedicated Clouds

Enterprise agreement lets companies license weights directly and run video generation on isolated clusters with full data controls.

Baseten and Runway formed a partnership on Wednesday to deploy Runway video models within dedicated cloud environments. Under the setup, corporate clients license weights for Gen-4.5 and Aleph 2.0 directly from the AI developer while Baseten provisions and manages inference on isolated hardware clusters.

The move targets a persistent bottleneck for Hollywood studios, advertising agencies, and corporate marketing teams. Individual creators typically use Runway Creative through a browser or send requests through shared public APIs. Organizations handling unreleased footage, character designs, and confidential product launches, however, routinely block workflows that pass sensitive files through public multi-tenant servers.

Direct licensing for Runway video models and managed hardware

The commercial structure splits intellectual property from operational compute. Runway controls the models and negotiates weight licensing directly with enterprise accounts. Baseten then deploys and maintains the underlying clusters through Baseten Dedicated Inference.

Everything runs in a single-tenant environment, meaning each company gets private silicon rather than sharing GPUs or memory with outside workloads.

The product lineup addresses two distinct production needs. Gen-4.5 generates video from text prompts or reference images, focusing on temporal physics and controlled camera movement. Aleph 2.0 functions as an in-context editor: it modifies lighting, framing, and visual assets across real or synthetic footage up to 30 seconds at 1080p resolution, cutting the need for reshoots. Companies can also run fine-tuning jobs on Baseten hardware to adapt model weights to internal style guides and proprietary assets.

Enterprise compliance and network isolation

To meet enterprise IT requirements, customer workloads run inside isolated virtual private clouds (VPCs). Traffic between corporate networks and the clusters travels over private endpoints using AWS PrivateLink or Google Cloud Private Service Connect.

According to Baseten documentation, the hosting platform holds SOC 2 Type II certification and HIPAA compliance. The company retains neither synchronous inference prompts nor rendered outputs on persistent disk. The dedicated offering also includes a formal service-level agreement (SLA) promising 99.99% system availability alongside dedicated GPU capacity.

Enterprise deployments and custom pricing

Runway already counts major names among its production users, including Amazon, Robinhood, Dolce & Gabbana, and Microsoft. Moving these models into dedicated clusters represents an effort to shift generative video from browser experimentation into governed production workflows.

Neither company disclosed standard contract pricing. Weight licensing through Runway and compute reservations through Baseten both require custom enterprise quotes. The providers also declined to specify default accelerator hardware, such as NVIDIA H100 or B200 GPUs, noting that actual chip allocations depend on the throughput and latency requirements negotiated for each deployment.