Comfy-Org · robinjhuang · Sep 9, 2026 · Sep 10, 2026 · wei-hai · Sep 10, 2026
diff --git a/development/api-development/sdks.mdx b/development/api-development/sdks.mdx
@@ -1,27 +1,27 @@
 ---
 title: "Comfy SDKs"
 description: "Official Python and TypeScript SDKs for running ComfyUI workflows from your own application"
 sidebarTitle: "SDK Overview"
 icon: "code"
 ---

 <Warning>
  **Beta.** The SDKs and the Comfy API v2 they call are at `0.1.x`. The shape of the API can still change before we lock it down. Now is the cheapest time to tell us it is wrong. See [Feedback](#feedback).
 </Warning>

 The Comfy SDKs let your application run ComfyUI workflows and get the results back. You submit a workflow, ComfyUI executes it, and you download the outputs. The same code runs against Comfy Cloud or against a ComfyUI instance you host yourself. Only the base URL changes.

 The SDKs are clients for the [Comfy API v2](/api-reference/v2/overview), a versioned HTTP API that we intend to support long term. New releases of ComfyUI will not break integrations built on it.

 Things people build this way:

 - Plugins that generate content inside another application, such as Blender or Krita
 - Consumer apps that run generation on behalf of their users
 - Batch pipelines, for example running one workflow over every frame of a video
 - Backend services that need many workflows in flight at once

 <Note>
  These SDKs drive ComfyUI **from the outside**. If you are writing custom nodes or frontend extensions that run **inside** ComfyUI, you want [Develop Custom Nodes](/custom-nodes/overview) instead. Those are a separate set of APIs.
 </Note>

 ## Install
@@ -44,12 +44,12 @@

 <CodeGroup>
 ```python Python
 from comfy_sdk import Comfy

 # Comfy Cloud
 client = Comfy(api_key="comfyui-...")

 wf = client.workflows.from_file("workflow_api.json")

 asset = client.assets.from_file("photo.png")
 wf.set_input("10", "image", asset)
@@ -119,14 +119,14 @@
 Works out of the box. Create an [API key](/development/api-development/getting-an-api-key) and pass it to the client.
 
 <Note>
-  API access requires a paid Comfy Cloud subscription. The free tier does not include it. How many jobs you can run at once depends on your tier. See [Parallel Execution](/development/deploy/cloud#parallel-execution-concurrent-jobs).
+  API access requires a paid Comfy Cloud subscription. The free tier does not include it. How many jobs you can run at once depends on your tier. See [Concurrency Limits](/development/deploy/cloud#concurrency-limits).
 </Note>
 
 ### Serverless deployment

 A workflow you deployed through the [developer platform](https://platform.comfy.org) gets its own endpoint. Point `COMFY_BASE_URL` at it and use your API key, exactly as with Comfy Cloud. Everything in this guide works the same way.

 Serverless deployments run one pinned workflow, so `get_workflow()` returns the executed graph (`format: "api"`).

 ### Your own ComfyUI

@@ -146,7 +146,7 @@

 <CodeGroup>
 ```python Python
 from comfy_sdk import Progress, Preview, OutputReady, StatusChange

 job = client.submit(wf)

@@ -154,7 +154,7 @@
    match event:
        case Progress() as p:
            print(f"{p.value:.0%} {p.message}")
        case Preview() as pv:
            image = pv.to_pil()
        case OutputReady() as o:
            o.output.to_file(f"partial/{o.output.name}")
@@ -188,7 +188,7 @@

 `result()` returns the finished job, or raises `JobFailed` with the node-level detail if execution failed. See the [SDK README](#reference) for your language for the full event catalog.

 The stream is a live feed, not a replayable log. It exists so you can render progress, not so you can rely on it for results. Polling the job is what is authoritative, and `run()`, `wait()`, and `result()` fall back to polling automatically. See [Design Notes](/development/api-development/sdks-design#poll-first-stream-for-progress) for why.

 ## Tracing an output back to its workflow

@@ -208,11 +208,11 @@

 The same id is on an asset fetched on its own, so a file you found later still leads back to its job. It is `None` (TypeScript: `undefined`) for an asset you uploaded, which has no producing job.

 From the job you can ask for the workflow behind it. This works for a job you did not submit in this process, rehydrated by id:

 <CodeGroup>
 ```python Python
 wf = job.get_workflow()

 if wf.format == "save":
    ...  # the workflow as authored, canvas layout and Note nodes intact
@@ -236,11 +236,11 @@
 | `format` | What you get | When |
 |---|---|---|
 | `save` | The authoring workflow at the version the job ran, with canvas layout and editor-only nodes such as Note | Jobs submitted from the Comfy Cloud editor, which pins a workflow version |
 | `api` | The executed graph. Editor-only constructs are gone and Get/Set nodes are expanded | Everything else, including every job submitted through these SDKs today |

 Jobs you submit through the SDK always return `api`, because v2 submission has no version-pinning fields yet. That will change; the discriminator is there so your code does not have to.

 ## What the SDKs cover today

 The first version does one thing properly: run a workflow and get the results back.

@@ -259,7 +259,7 @@

 ## Reference

 The SDK READMEs are the full reference for each language, including auth, assets, errors, and the low-level escape hatches.

 <CardGroup cols={2}>
  <Card title="Python SDK" icon="python" href="https://github.com/Comfy-Org/comfy-python-sdk">
@@ -269,7 +269,7 @@
    <code>@comfyorg/sdk</code> on npm. Typed, async, with a low-level client.
  </Card>
  <Card title="Comfy API v2 Reference" icon="code" href="/api-reference/v2/overview">
    The HTTP API underneath both SDKs. Use it directly from any language.
  </Card>
  <Card title="Design Notes" icon="compass" href="/development/api-development/sdks-design">
    Why this API exists, how it relates to the existing ComfyUI APIs, and what comes next.
@@ -282,4 +282,4 @@

 So tell us what is awkward, what you expected to find and did not, and what you ended up working around. The `#developer-platform` channel in [our Discord](https://discord.com/invite/comfyorg) is the place for it.

 If you want a first-party SDK in another language, say so there. Both SDKs sit on the same documented HTTP contract, so any language can talk to the API today, but we would rather know where the demand is.
diff --git a/development/cloud/api-reference.mdx b/development/cloud/api-reference.mdx
@@ -185,7 +185,7 @@
 ### Upload Mask

 <Note>
  The `subfolder` parameter is accepted for API compatibility but ignored in cloud storage. All files are stored in a flat, content-addressed namespace.
 </Note>

 <CodeGroup>
@@ -261,7 +261,7 @@
 Submit a workflow for execution.
 
 <Info>
-  **Concurrent submissions supported:** Depending on your subscription tier, you can submit multiple workflows without waiting for previous jobs to complete. Jobs run in parallel up to your tier's limit. Additional jobs queue automatically. See [Parallel Execution](/development/deploy/cloud#parallel-execution-concurrent-jobs) for details and concurrency limits.
+  **Concurrent submissions supported:** Depending on your subscription tier, you can submit multiple workflows without waiting for previous jobs to complete. Jobs run in parallel up to your tier's limit. Additional jobs queue automatically. See [Concurrency Limits](/development/deploy/cloud#concurrency-limits) for details.
 </Info>
 
 ### Submit Workflow
@@ -411,8 +411,8 @@
 ```typescript TypeScript
 function setWorkflowInput(
  workflow: Record<string, any>,
  nodeId: string,
  inputName: string,
  value: any
 ): Record<string, any> {
  if (workflow[nodeId]) {
@@ -513,7 +513,7 @@
    | Offset | Size | Field | Description |
    |--------|------|-------|-------------|
    | 0 | 4 bytes | `type` | `0x00000003` |
    | 4 | 4 bytes | `node_id_len` | Length of node_id string |
    | 8 | N bytes | `node_id` | UTF-8 encoded node ID |
    | 8+N | variable | `text` | UTF-8 encoded progress text |
  </Tab>
@@ -579,7 +579,7 @@
 }

 async function waitForCompletion(
  promptId: string,
  timeout: number = 300000
 ): Promise<Record<string, any>> {
  const wsUrl = `wss://cloud.comfy.org/ws?clientId=${crypto.randomUUID()}&token=${API_KEY}`;
@@ -626,10 +626,10 @@

 async function downloadOutputs(
  outputs: Record<string, any>,
  outputDir: string
 ): Promise<void> {
  await mkdir(outputDir, { recursive: true });
  for (const nodeOutputs of Object.values(outputs)) {
    for (const key of ["images", "video", "audio"]) {
      for (const fileInfo of (nodeOutputs as any)[key] ?? []) {
        const params = new URLSearchParams({
@@ -643,9 +643,9 @@
          redirect: "manual",
        });
        if (response.status !== 302) throw new Error(`HTTP ${response.status}`);
        const signedUrl = response.headers.get("location")!;
        // Fetch from signed URL without auth headers
        const fileResponse = await fetch(signedUrl);
        if (!fileResponse.ok) throw new Error(`HTTP ${fileResponse.status}`);

        const path = `${outputDir}/${fileInfo.filename}`;

diff --git a/development/comfy-router/quickstart.mdx b/development/comfy-router/quickstart.mdx
@@ -10,7 +10,7 @@ Comfy Router lets you call partner models through `https://api.comfy.org` with o
 
 <Steps>
   <Step title="Create an API key">
-    Create a key in [your Comfy workspace](https://platform.comfy.org/profile/api-keys). In a Bash-compatible terminal, set:
+    Create a key in the [Developer Platform](https://platform.comfy.org/profile/api-keys). In a Bash-compatible terminal, set:
 
     ```bash
     export COMFY_API_KEY="comfyui-..."

diff --git a/development/deploy/cloud.mdx b/development/deploy/cloud.mdx
@@ -5,7 +5,7 @@
 ---


 Comfy Cloud is a stateful, interactive, fully managed version of ComfyUI. You build and run workflows in the browser as you would on a local install, while Comfy manages the models, the GPUs, and everything in between. You can also call Comfy Cloud through the API. Comfy Cloud provides a managed endpoint, so there is no environment to build and no deployment to create.

 Unlike a [serverless deployment](/development/serverless/overview), which serves one pinned environment that you build and release, Comfy Cloud serves a full, managed ComfyUI that is always on.

@@ -19,7 +19,7 @@
 
 <Steps>
   <Step title="Create an API key">
-    Log in at [platform.comfy.org](https://platform.comfy.org) and create a key. See [Getting an API Key](/development/api-development/getting-an-api-key) for step-by-step instructions.
+    Log in to the [Developer Platform](https://platform.comfy.org) and create a key. See [Getting an API Key](/development/api-development/getting-an-api-key) for step-by-step instructions.
   </Step>
   <Step title="Install the SDK">
     <CodeGroup>
@@ -38,7 +38,7 @@
  <Step title="Run it">
    <CodeGroup>
    ```python Python
    from comfy_sdk import Comfy

    client = Comfy(api_key="comfyui-...")  # Comfy Cloud is the default target

@@ -65,22 +65,24 @@
  </Step>
 </Steps>

 See [Running Workflows](/development/run-workflows/overview) for how the same code moves between Comfy Cloud, serverless deployments, and self-hosted deployments.

 ## Credits and Usage
 
 API requests draw from the same monthly credit allocation as the Comfy Cloud web UI. There is no separate API credit pool. Each tier's included credits, top-up options, and per-workflow runtime caps apply to API jobs in exactly the same way as UI jobs. See the [pricing page](https://www.comfy.org/cloud/pricing?utm_source=docs&utm_campaign=cloud-api) for the monthly credit amounts on each tier. If you run out of credits mid-month, top-ups can be purchased from your account dashboard.
 
-## Parallel Execution (Concurrent Jobs)
+## Concurrency Limits
 
-API users can submit multiple workflows concurrently without waiting for previous jobs to complete. Submission returns as soon as the job is accepted, so you can keep several in flight. The dispatcher runs them in parallel up to your subscription tier's limit; check [platform.comfy.org](https://platform.comfy.org) for your current concurrency.
+Running multiple jobs in parallel is only supported when you submit workflows through the API. API users can submit multiple workflows concurrently without waiting for previous jobs to complete. Submission returns as soon as the job is accepted, so you can keep several in flight. The dispatcher runs them in parallel up to your subscription tier's limit; check [platform.comfy.org](https://platform.comfy.org) for your current concurrency.
 
 Jobs submitted beyond your concurrency limit queue normally and execute automatically as slots free up. If the queue itself is full, the SDKs retry for you within a bounded budget before raising `QueueFull`.
 
 <Info>
-  Parallel execution is currently available via the API only. See [pricing plans](https://www.comfy.org/cloud/pricing?utm_source=docs&utm_campaign=cloud-api) for subscription details.
+  The Comfy Cloud UI does not support running more than one job per person. Workflows queued in the editor run one at a time, in order. Only API submissions run in parallel. See [pricing plans](https://www.comfy.org/cloud/pricing?utm_source=docs&utm_campaign=cloud-api) for subscription details.
 </Info>
 
+On Team and Enterprise plans, concurrency is per member: everyone in the workspace has their own concurrency limit, so one member's running jobs never block another's.
+
 ## Existing v1 integrations
 
 The deprecated [v1 Cloud API](/development/cloud/overview) also runs against Comfy Cloud. Use it only for existing integrations or for Cloud capabilities that [Comfy API v2](/api-reference/v2/overview) does not expose yet. Runnable examples are in the [Cloud API Reference](/development/cloud/api-reference) (deprecated).
diff --git a/get_started/cloud.mdx b/get_started/cloud.mdx
@@ -1,6 +1,6 @@
 ---
 title: "Comfy Cloud - Official ComfyUI Cloud Platform Powered by RTX 6000 Pro"
 description: "Run ComfyUI workflows online with Comfy Cloud, the official cloud platform powered by NVIDIA RTX 6000 Pro GPUs. No local installation required."
 sidebarTitle: "Comfy Cloud"
 icon: "cloud"
 ---
@@ -21,7 +21,7 @@
  </Card>

  <Card title="Powerful GPUs" icon="microchip">
    Run workflows fast on our powerful server GPUs without needing your own hardware
  </Card>

  <Card title="Always up-to-date" icon="arrows-rotate">
@@ -44,7 +44,7 @@
 | **GPU** | Powerful Blackwell RTX 6000 Pros | Bring your own GPU |
 | **Technical Knowledge** | No technical knowledge required. | While desktop and portable give you easy ways to get started, you'll need to troubleshoot custom node installations and local installation issues. |
 | **Custom Nodes** | Use pre-installed custom nodes and never worry about compatibility issues. | Install any custom node you want, but you'll need to manage it yourself. |
 | **Models** | Use pre-installed models. Import LoRA models from Civitai. Import models from Hugging Face (coming soon). | Use any models you want, but you'll need to download them first. |
 | **Notable Differences** | Easy to onboard your team | Works offline, infinitely customizable |
 | **Get started** | [Run ComfyUI Cloud](https://comfy.org/cloud) | [Install ComfyUI locally](/installation/system_requirements) |

@@ -55,6 +55,14 @@
   View pricing and subscription options for Comfy Cloud
 </Card>
 
+### Concurrency limits
+
+The Comfy Cloud UI does not support running more than one workflow at a time. Each person can run one workflow at a time in the editor; additional workflows wait in the queue and run one after another.
+
+Running multiple jobs in parallel is only supported when you submit workflows through the API, which requires a paid subscription. See [Concurrency Limits](/development/deploy/cloud#concurrency-limits) in the developer docs for details.
+
+On Team and Enterprise plans, concurrency is per member: everyone in the workspace has their own concurrency, so one member's running workflows never block another's.
+
 
 ## How to use ComfyUI Cloud
 
@@ -78,7 +86,7 @@
  </Step>

  <Step title="Run your workflow">
    If everything is correct, click the "Run" icon or use the shortcut "Ctrl + Enter" to run the workflow.
    ![Run workflow](/images/cloud/workflow_run_workflow.webp)
  </Step>