Save 4 hours! We auto-setup your workflow! Free!

Drop your workflow.json — we handle every dependency, custom node, and model. Just open the link and run.

Auto-Setup Workflow Json (Free) Now!
ComfyUI > Nodes > ComfyUI-API-Toolkit

ComfyUI Extension: ComfyUI-API-Toolkit

Repo Name

ComfyUI-API-Toolkit

Author
ixmxamar (Account age: 0 days)
Nodes
View all nodes(1)
Latest Updated
2026-10-02
Github Stars
0.0K

How to Install ComfyUI-API-Toolkit

Install this extension via the ComfyUI Manager by searching for ComfyUI-API-Toolkit
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter ComfyUI-API-Toolkit in the search bar
After installation, click the Restart button to restart ComfyUI. Then, manually refresh your browser to clear the cache and access the updated list of nodes.

Visit ComfyUI Online for ready-to-use ComfyUI environment

  • Free trial available
  • 16GB VRAM to 80GB VRAM GPU machines
  • 400+ preloaded models/nodes
  • Freedom to upload custom models/nodes
  • 200+ ready-to-run workflows
  • 100% private workspace with up to 200GB storage
  • Dedicated Support

Run ComfyUI Online

ComfyUI-API-Toolkit Description

ComfyUI-API-Toolkit connects ComfyUI workflows to Kling AI, ElevenLabs, and Google Gemini to generate and edit video, images, speech, music, and text, with a pixel-art resizing utility.

ComfyUI-API-Toolkit Introduction

ComfyUI-API-Toolkit brings four collections of nodes into one ComfyUI extension: Kling AI for video, images, and audio; ElevenLabs for voices, sound, music, and transcription; Google Gemini for text, images, audio, video, and embeddings; and Utils for pixel-art resizing. The version described in the original README is v1.7.0, with 78 nodes in total.

If you create work across several media, the toolkit lets you connect those steps in a ComfyUI workflow. For example, you could develop a prompt with Gemini, generate a video with Kling, and create a voiceover with ElevenLabs. You can use only the services relevant to your project; an unavailable service does not prevent the other service collections from loading.

The toolkit connects to external AI services. You need an API key for each service you choose to use, and the models available to you can depend on your account.

How ComfyUI-API-Toolkit Works

Think of each node as a step in a creative process. A node takes inputs, performs a task, and passes its result to another node. A text-generation node might help you shape a scene description; a video-generation node can then use that description to create footage.

The toolkit groups nodes by service and purpose:

  • Configuration nodes provide a service’s API key and, where applicable, model or safety choices.
  • Creation nodes request an image, video, voice, piece of music, or other result.
  • Utility nodes help prepare, upload, load, or save material used elsewhere in the workflow.

API-based generation sends a request to the selected service. Some requests may take time or encounter temporary rate limits. The toolkit includes retry behavior that waits progressively longer after transient failures. Its API nodes also request a fresh result each time you queue the workflow, instead of simply reusing ComfyUI’s cached output.

You can hover over a node’s inputs to read tooltips explaining its settings. API-key inputs are password-masked, and the authentication nodes can also read keys from environment variables. Masking helps keep keys out of screenshots, but you should still take care when sharing workflow files or account credentials.

ComfyUI-API-Toolkit Features

Kling AI: Video, images, and audio

The 38 Kling nodes cover text-to-video, image-to-video, video extension, lip sync, motion control, avatars, image creation and editing tasks, virtual try-on, audio generation, text-to-speech, voice cloning, effects, and upscaling. Other nodes handle assets, camera controls, file loading, and saving.

For a short moving scene, you might start with Kling Text to Video. If you already have a character image and want to animate it, Kling Image to Video is the more relevant starting point. Kling Lip Sync can help match speech to a video, while Kling AI Upscale is intended for improving output resolution.

Kling Text to Video, Image to Video, and Video Omni offer 3–15-second durations, a 4K mode, and multi-shot storyboards. For a storyboard, enter one shot per line in seconds|prompt form. Duration changes the length of the requested clip; separate shot descriptions let you plan a sequence rather than describe everything in one prompt. Kling authentication accepts either a single API key, which is required for Kling 3.0 Turbo, or an access-key and secret-key pair.

ElevenLabs: Voices, dialogue, and sound

The 18 ElevenLabs nodes support text-to-speech, speech-to-speech, voice cloning, voice design and remixing, sound effects, audio isolation, transcription, dialogue, and music generation. There are also nodes for selecting or fetching voices and checking account information.

Use Text to Speech to turn written narration into spoken audio. TTS with Timestamps is useful when timing matters, while Text to Dialogue supports scenes with multiple speakers. Speech to Text works in the opposite direction, turning audio into a transcript. Voice selection changes who the speech sounds like; changing the text or dialogue changes what is said.

Google Gemini: Ideas and multimodal creation

The 21 Gemini nodes include configuration and model-selection tools alongside text, image, audio, video, and embedding nodes.

  • Text tools generate writing, refine prompts, maintain multi-turn chats, or produce structured output.
  • Image tools analyze an image, generate a new one, or edit, inpaint, and outpaint an existing image.
  • Audio tools create speech or music.
  • Video tools generate video, including options for first- and last-frame input, or use Gemini Omni for generation, editing, and extension.
  • Text Embeddings produces numerical representations of text for workflows that need to compare or organize text by meaning.

For example, use Gemini Vision Analysis to describe a reference image before writing a prompt, Gemini Image Edit to change an existing image, or Gemini Outpaint when you want to expand beyond its current boundaries. The Gemini Safety Settings and Gemini Thinking Config nodes provide additional configuration. Gemini List Available Models checks which models your API key can actually access, and Gemini Token Counter helps you assess the size of a text request.

Utils: Pixel Art Resize

Pixel Art Resize turns images into pixel art. It includes palette locking across frames, which can help an animation maintain consistent colors, and Floyd–Steinberg dithering, which can create the appearance of additional shades using a limited palette.

Workflow conveniences

Services are discovered automatically, and each service has optional dependencies. If one service cannot load, the others can remain available. The toolkit also creates missing output directories when saving, and its cloud uploader tries catbox before falling back to tmpfiles. Bear in mind that uploading an asset sends it to an external hosting service.

ComfyUI-API-Toolkit Models

The toolkit does not contain its own image, video, or voice model. Its nodes give you access to models offered by the connected services. The original README identifies the following choices; the models available to your account may differ, so use a service’s model-selection tools where provided.

ServiceModels described in the original READMEWhen to consider them
Kling AIKling 3.0 Turbo options for text-to-video and image-to-videoWhen using the corresponding Turbo video nodes; these require a Kling API key.
ElevenLabsEleven v4 and v3; Scribe v2 and its medical variant; Music v1, v2, and v2.5Choose according to whether your task is speech creation, transcription, or music generation.
Google Gemini and GemmaGemini 3.x and 2.5 text/multimodal options, latest aliases, and Gemma-family optionsFor writing, conversation, prompt refinement, vision, and other supported text or multimodal tasks.
Google image modelsFour Nano Banana image-generation options and Imagen 4 Ultra, Standard, and FastFor image generation; choose a model supported by the image node and your account.
Google audio modelsFive TTS models with more than 30 prebuilt voices; three Lyria music modelsFor spoken audio or music. Voice choice affects the sound of generated speech.
Google video modelsVeo 3.1 standard, fast, and lite variants, plus Gemini Omni FlashFor video generation or the generation and editing tasks supported by Gemini Omni.
Google embedding modelsThree embedding optionsFor workflows that use text embeddings rather than directly creating visible artwork.

The original README also mentions specialized Google options for Robotics ER 2, Deep Research, and TTS previews, but does not give artist-facing workflows for them. It notes that Google scheduled Imagen 4 models for shutdown on August 17, 2026; check current availability with Gemini List Available Models rather than assuming an older model choice will work.

What's New with ComfyUI-API-Toolkit

The provided README describes v1.7.0 but does not include a version-by-version changelog. It highlights these capabilities in that version:

  • Four service collections and 78 nodes: one toolkit covers a range of visual, audio, and text tasks.
  • Expanded Gemini coverage: the Gemini collection goes beyond image nodes to include text, vision, image editing, speech, music, video, and embeddings.
  • Service-by-service loading: an unavailable optional dependency for one service does not have to stop the others from working.
  • Reliability and usability features: API retries, fresh results when workflows are queued, masked key inputs, parameter tooltips, and automatic output-directory creation.

These are capabilities described for v1.7.0, not a claim that each was introduced in that release.

Troubleshooting ComfyUI-API-Toolkit

A service’s nodes do not appear in ComfyUI

  1. Check whether nodes from the other toolkit services appear. The extension allows services to load independently.
  2. Confirm that the unavailable service’s optional dependency is present in the ComfyUI environment you are using. The original README identifies requests and PyJWT for Kling, requests, soundfile, and scipy for ElevenLabs, and google-genai>=2.3.0 for Gemini.
  3. Restart ComfyUI after correcting the dependency, then check whether the nodes appear.

A node reports an authentication error

  1. Confirm that you supplied a key for the service used by that node.
  2. For Kling, check whether the node expects a single API key or an access-key and secret-key pair. Kling 3.0 Turbo requires the single API key.
  3. If you rely on environment variables, check the relevant name: KLING_API_KEY or KLING_ACCESS_KEY and KLING_SECRET_KEY; ELEVENLABS_API_KEY; or GEMINI_API_KEY.
  4. Check that the key has access to the model or feature you selected. For Gemini, Gemini List Available Models can help you see what your key can access.

A Gemini model option does not work

  1. Run Gemini List Available Models with your API key.
  2. Select an accessible model appropriate to the node’s task.
  3. If you are using Gemini 3.8 TTS, Lyria, or Gemini Omni, check that the installed google-genai package is version 2.3.0 or newer, as specified in the original README.

A request is delayed or fails temporarily

API services can experience transient errors or rate limits. The toolkit retries those failures with increasing waits. Give the request time to complete; if it continues to fail, check your service account, selected model, and request inputs before queuing it again.

Frequently Asked Questions

Do I need keys for all three AI services?
No. You need keys only for the services you use.

Will a missing service stop every toolkit node from loading?
No. The toolkit is designed so that the other service collections can continue loading.

Why did I get a different result when I queued the same workflow again?
The API nodes request fresh results on each queue rather than relying on ComfyUI’s cached output. Generated results can vary between requests.

Does this extension include AI models on my computer?
No. The listed creation nodes connect to external AI services. The Pixel Art Resize utility is a separate image-processing node within the toolkit.

Can I share a screenshot showing my authentication node?
API-key fields are password-masked to reduce accidental exposure in screenshots. Still review anything you share, especially workflow files and other account details.

Learn More about ComfyUI-API-Toolkit

The ComfyUI-API-Toolkit repository is the project link provided in the original README. Within ComfyUI, hover over a node input for its parameter tooltip, and use Gemini List Available Models when you need to check model access for your Google API key.

The author also lists separate packages for artists who only need one service: ComfyUI-Kling-Direct, ComfyUI-ElevenLabs-Pro, and ComfyUI-NanoBanana2.

ComfyUI-API-Toolkit Related Nodes

RunComfy
Copyright 2025 RunComfy. All Rights Reserved.

RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.

ComfyUI-API-Toolkit detailed guide | ComfyUI