More and more people let AI agents work in Blender for them, using tools like Claude Code, Codex, or Cursor. The agent works through Blender MCP or writes and runs Python scripts. It can set up a scene quickly. Rendering it is another matter, especially for final-quality stills, animations, or dozens of variations.
That part can run on RenderStreet. Your agent uploads the project, starts a render through our API, and carries on with its work. When the frames are ready, you get a notification and the agent can pick them up.
Where rendering gets in the way
Agents handle a lot of the technical work in Blender well: blocking out a scene, assigning materials, making variations, checking a file before it goes out. Rendering is usually where things slow down.
A final Cycles render or an animation keeps your computer busy, and that's the same computer running Blender and the agent. If the agent works in a loop (render, look at the result, fix, render again), every round waits for the render to finish. Long renders also don't fit well into AI tools, which expect a quick reply and can time out. And a live Blender session holds one scene at a time, so running several agents in parallel means moving to background scripts anyway.
Sending renders to RenderStreet as background jobs solves most of this. Your machine stays free and the agent doesn't have to wait.
How it works
-
Build the scene on your machine
The agent creates or edits the scene in your local Blender, with Blender MCP or a script run through
blender -b. Quick checks in the viewport or at low samples stay local. -
Upload the project
The agent uploads the project folder over FTP, with its textures, linked files, and caches.
-
Start the render
An API call starts the job with your render settings. The response comes back right away with a job ID.
-
Keep working
While the job renders, the agent can do other things. It can ask the API for progress at any time, and a webhook tells your system when the job is done.
-
Get the frames
The finished frames are downloaded automatically. If something needs fixing, the agent changes the scene and sends it again.
Who is it for
Developers building asset pipelines with agents
If you run Claude Code or Codex with Blender in the background, maybe with several agents at once, each job you send renders on its own servers. A batch of assets or turntables comes back together instead of rendering one by one on your machine.
Filmmakers and previs artists working with AI
When an agent helps you iterate on shots, fast GPU renders on NVIDIA L40s shorten each round, and the final Cycles animation can render while you work on the next shot.
Archviz and product studios
When you need every camera angle, material, or product variant rendered, the agent can set up the variations and we render them in parallel. For very large projects, have a look at our batch rendering page.
Technical directors
Many TDs use agents for repetitive fixes, scene checks, and pipeline scripts. Your projects render on isolated servers, and Python scripts and add-ons can be enabled for your project when you need them.
Teams generating synthetic training data
If you produce datasets with scripted changes in lighting, cameras, and materials, parallel jobs and automatic downloads keep long dataset runs going. More on this on our batch rendering page.
Teams building AI tools on top of Blender
If you're building a product where users describe a scene and get images or video back, RenderStreet can be your rendering backend. Your service sends the jobs through the API and gets the frames back automatically, so you don't need to run your own render servers.
Why RenderStreet
- Jobs run in parallel. Send as many jobs as you need. We allocate servers automatically based on the workload.
- GPU or CPU, at regular prices. NVIDIA L40s with 48 GB VRAM at $4.49 per server-hour, or CPU servers at $3.00 per server-hour, billed by the minute. There's no extra fee for using the API.
- Your files remain private and secure, whether rendered on our bare-metal machines or in the cloud infrastructure we integrate.
- Any Blender version. Blender 5.3 and every version back to 2.75. New releases are usually available within days.
- The whole project, not just the .blend. Upload everything the scene depends on. Most add-ons work as they are, and we send setup instructions for the ones that need it.
- A track record. We've been rendering since 2012: more than 135 million frames for artists and studios in 112 countries, with a 99% success rate.
What's next for the API
We're working on a new public API, with self-serve access and public documentation. When it launches, RenderStreet One subscribers will be able to send jobs through the API too. Until then, request access below. We'll set up your account and help you connect your workflow.
Frequently asked questions
Which AI tools does it work with?
Any agent or AI-based tool that can make HTTP requests can use it, because it's a standard REST API. That includes Claude Code, OpenAI Codex, Cursor, Gemini CLI, and agents you build yourself.
Can I use it with Blender MCP?
Yes, the two work well together. Blender MCP tools, including the official Blender Lab server behind the Claude Blender connector, let the agent build and edit the scene in your local Blender. Once the scene is ready, the agent uploads it and starts the render through our API, so the heavy rendering happens on our servers.
Is my scene safe?
Your project files render on isolated servers that we control. Python script execution is off by default, and we can enable it for your project if your pipeline needs it.
How is API rendering billed?
There's no extra fee for using the API. API jobs are billed at our regular prices, per server-hour and by the minute: $3.00 on CPU servers and $4.49 on NVIDIA L40s GPUs with 48 GB VRAM. If you render large volumes regularly, get in touch and we'll talk about volume pricing.
Can my agent run several jobs at once?
Yes. One account can run many jobs at the same time, and we allocate servers automatically based on the workload. Variations, camera angles, or assets render in parallel instead of waiting in line on one machine.
How do I get API access?
Use the Request API access button and tell us a bit about your project and the tools you use. We'll set up your account and send you the API documentation. Access is open to individual artists, developers and studios alike.
Connect your agent to RenderStreet
Tell us a bit about your workflow and the tools you use. We'll set up your API access and help you get your first jobs rendering.