Define the build
Start from a known-good environment or import a snapshot of a setup you already run.
Govern the models, custom nodes, and dependencies your team runs. Create a managed distribution of ComfyUI and deploy the same build anywhere, local or serverless cloud.
MANAGED BUILDS
Define the ComfyUI release, custom nodes, models, and dependencies your team is approved to run. Ship the same versioned build wherever the work happens.
COMFYUI RELEASE
PYTHON · PYTORCH · CUDA
CUSTOM NODES
MODELS
v0.3.41 · py3.12 · torch 2.6
4 nodes · 3 models · pinned
Start from a known-good environment or import a snapshot of a setup you already run.
Pin the ComfyUI release, Python and CUDA versions, custom nodes, models, and dependencies.
Cut a new version when you decide. Move the team together without changing work already in flight.
Assign the build through Comfy Desktop or your GPU servers. Deploy endpoints through the Developer Platform.

The graph stays flexible. The environment around it becomes an approved, reviewable operating model for the organization.
For local deployments, your Comfy Workflows and outputs stay where they run. Comfy does not train on customer data.
Include internal node packs and fine-tuned models inside a team-controlled build.
Gate build assignment through company identity and use supported provider keys with existing contracts.
Define the model, partner, authentication, release, retention, and export events your organization needs.
DPA and enterprise agreement available. US processing on Google Cloud. 99.5% workflow-execution uptime SLA. SOC 2 Type II audit in progress.
Builds is self-serve for packaging and testing your own environment. Managed Builds adds team sharing and enterprise governance.
A Managed Build is an approved, reproducible ComfyUI environment. It pins the ComfyUI version, custom nodes, models, and dependencies your team needs so the same setup can be assigned and updated across machines.
Managed Builds run on your workstations, studio GPU servers, or customer cloud infrastructure. Your Comfy Workflows stay on the machines where they run.
Yes. Teams can include private fine-tuned models, govern which models and partner providers are available in a build, and use supported provider keys with their existing contracts.
Administrators choose which custom nodes belong in each build, then pin them with the rest of the environment for controlled rollout.
Company identity and SSO can gate access to approved builds. Teams receive the environments assigned to them instead of configuring every machine independently.
Track usage by model and partner on the Developer Platform. Authentication, build-assignment, release, retention, and export requirements are scoped explicitly during enterprise security review.
Request a demo and bring one known-good environment. Comfy engineers will map the current setup, deployment targets, governance requirements, onboarding plan, and path to the first approved build.
Builders, not advisors. Forward Deployed Creatives build production Comfy Workflows inside approved environments, alongside the teams that will own and extend them.