Baseten Adds Runway’s Gen-4.5 and Aleph 2.0 to Dedicated Inference

Enterprises can now deploy Runway’s Gen-4.5 and Aleph 2.0 video generation and editing models, including fine-tuned versions, in secure, single-tenant environments on Baseten, according to Baseten’s October 14, 2026 announcement. The models are available through Baseten Dedicated Inference, with model weights licensed directly from Runway.

According to Baseten, Runway builds models that generate and edit cinematic video using text, images, audio, and existing footage. Runway already offers Gen-4.5 and Aleph 2.0 through Runway Creative, its platform for generating and editing video, images, and audio, and through Runway Dev, its API platform for developers. Baseten said the partnership adds a new path: Runway is making its proprietary video models available for dedicated deployment on Baseten.

Gen-4.5 Generation and Aleph 2.0 Editing

Gen-4.5 turns text and image inputs into video, with support for detailed scene direction, camera movement, and a range of visual styles. Baseten lists uses including visualizing a storyboard, animating a product image, creating cinematic B-roll, and adapting a campaign concept into multiple variations.

Baseten’s model library describes Gen-4.5 as a video generation model developed by Runway that delivers cinematic motion, precise prompt adherence, realistic physics, and consistent visual quality, with strong temporal consistency across photorealistic and stylized aesthetics. The library entry lists text and image inputs, video output, and a proprietary Runway license.

Aleph 2.0 modifies generated or real footage. Starting from a prompt or an edited keyframe, users can change a specific element or restyle a scene, then carry that edit across relevant shots while preserving the surrounding details. The model supports multi-scene videos up to 30 seconds at 1080p.

Baseten’s library describes Aleph 2.0 as an in-context video editing model that turns existing footage into new variations without reshoots or extensive post-production. According to the entry, it can modify products, clothing, hairstyles, backgrounds, seasons, framing, and visual styles across multiple shots, and it lists text, image, and video inputs, video output, and a proprietary Runway license.

Licensing, Isolation and Security

Customers license Runway model weights directly from Runway and deploy them in a secure, single-tenant environment on Baseten. Each deployment runs in an isolated workload plane with dedicated GPU resources, while Baseten manages the control plane and underlying infrastructure. Runway and Baseten support each deployment together, from model customization through production serving.

Under the arrangement, compute and scheduling are isolated to the customer organization. Baseten’s access to Runway’s model weights is governed by a distribution agreement, and deployments run on infrastructure covered by Baseten’s SOC 2 Type II and HIPAA compliance programs.

Baseten’s single-tenant documentation adds architectural detail. It states that single-tenant runs inference workloads in an isolated VPC in Baseten Cloud, with compute and networking provisioned and operated for one organization, and that no other customer’s workloads run on that infrastructure. According to the documentation, Baseten does not store synchronous inference inputs or outputs, models can pull weights from a private source, and weights can be cached in-cluster to accelerate cold starts. On request, model replicas and supporting services can run without direct internet access, with outbound traffic limited to an approved allow-list, and endpoints can be exposed through private networking services such as AWS PrivateLink or Google Cloud Private Service Connect so traffic stays off the public internet. The documentation states that single-tenant supports the same compliance programs as Baseten Cloud, covering SOC 2 Type II, HIPAA, and GDPR, and is available with custom pricing on the Enterprise plan.

Baseten’s Dedicated Inference product page states that dedicated deployments are single-tenant, can be region-locked, and are HIPAA compliant and SOC 2 Type II certified on Baseten Cloud. Customers can deploy on Baseten Cloud, self-host, or use Baseten Hybrid, with autoscaling across nodes, clusters, clouds, and regions.

Fine-Tuning, Reserved Capacity and Uptime

Customers can fine-tune Runway models on their own data so outputs reflect their brand, canon, and content guidelines. The direct license with Runway governs that work, while Baseten provides the infrastructure to run the fine-tuning and deploy the resulting model in a dedicated environment.

Deployments include reserved GPU capacity for video-generation workloads, with autoscaling to handle spikes in demand. Under the single-tenant model, Baseten procures and operates capacity for each dedicated workload plane, and customers work with the Baseten team to select regions, GPU types, and capacity commitments that meet workload requirements.

Baseten said Amazon, Robinhood, Dolce & Gabbana, and Microsoft are among the enterprise creative teams already using Runway in production. The company said it backs dedicated deployments with a 99.99% uptime SLA, and both Runway models are listed in Baseten’s model library for deployment inquiries.