Why NodeTool

The problems NodeTool solves for creative teams working with AI models.

NodeTool is the open creative AI workspace — every major model, your keys, one canvas.

NodeTool replaces the chatbox with a node canvas where image, video, audio, and LLM models run side by side. Bring your keys, or run everything locally. Open source, AGPL-3.0.

The problem

Creative AI is a tab graveyard:

  • Every model lives behind its own UI. Switching between Flux, ElevenLabs, Sora, and a chatbot means losing context every time. Outputs don’t compose.
  • SaaS canvases tax every token. Hosted creative tools mark up provider credits 2–5x. You pay for the same OpenAI call twice.
  • Local-only tools are model-narrow. ComfyUI nails diffusion internals but treats LLMs, agents, and cloud APIs as second-class.
  • Workflows aren’t portable. Prompts in chat history, settings in screenshots, pipelines in your head. Nothing is reproducible.
  • Privacy is a yes/no toggle. Either everything goes to a vendor, or nothing leaves your machine. No mixed mode.

How NodeTool solves this

One canvas, every model, your keys:

One canvas for every modality. Wire Flux to GPT-5 to ElevenLabs to Wan in a single graph. Outputs flow as typed edges — image, audio, text, embeddings — not pasted strings.

BYOK to every provider. OpenAI, Anthropic, Gemini, Replicate, FAL, Kie, ElevenLabs, MiniMax, HuggingFace. Pay them directly. No credit markup, no provider lock-in.

Local + cloud, mixed. Run Llama on MLX, route image gen to FAL, send audio to ElevenLabs — in one workflow. Toggle per node.

Workflows as files. Save, share, version. Ship a workflow as a Mini-App with a one-click hide-the-graph mode.

Agents that drive pipelines. Built-in planning, tool calling, streaming. Drop an agent node into any graph.

Open source, no markup. AGPL-3.0. Cloud edition hosts the same code in this repo. Self-host the Docker images any time.

Feature Comparison

Feature NodeTool Weavy ComfyUI
Category Open creative AI workspace Closed SaaS creative canvas Diffusion-focused node editor
License AGPL-3.0 (open source) Proprietary SaaS GPL-3.0 (open source)
Runs on your machine ✅ Mac, Windows, Linux desktop ❌ Browser-only, hosted ✅ Local-first
Bring your own keys (BYOK) ✅ FAL, Kie, OpenAI, Anthropic, Gemini, Replicate, Atlas, ElevenLabs, MiniMax ❌ Credits only, provider markup ⚠️ Via custom nodes for cloud APIs
Pricing model Pay providers directly, no markup Proprietary credits Free (you pay your own GPU/API)
Model coverage Image, video, audio, text, TTS, ASR — local + cloud Image, video, audio — cloud only Diffusion (image/video) — local
Image generation Local: FLUX, Qwen Image · API: FAL, Kie, Replicate, OpenAI, Gemini Cloud: FLUX, Seedance, Ideogram, etc. Deep control over diffusion internals
Video generation Local: Wan · API: FAL, Kie, Sora, Veo, Kling Cloud: Kling, Veo, Runway, etc. Local diffusion video (AnimateDiff, etc.)
Audio & music Local: MusicGen, AudioLDM, Stable Audio · API: Kie, ElevenLabs, MiniMax Cloud: Suno, ElevenLabs, etc. ⚠️ Via custom nodes
TTS / ASR Local: Kokoro, Sesame, Whisper · API: OpenAI, ElevenLabs Cloud only ⚠️ Via custom nodes
LLMs & agents Built-in agent nodes, tool calling, streaming, Ollama, MLX Limited LLM nodes ⚠️ Via custom nodes
Diffusion control Standard parameters ❌ Hidden behind presets ✅ Latents, VAE, samplers, ControlNet
RAG / vector search ✅ Local SQLite-vec, plus Pinecone & Supabase pgvector ❌ ❌
Mini-apps from workflows ✅ Turn a graph into a simple UI ⚠️ Share-as-template ❌
Real-time streaming ✅ Token-by-token, live progress ✅ Live preview ❌ Queue-based execution
Source available ✅ Full source on GitHub ❌ ✅ Full source on GitHub

When to pick each

NodeTool — every modality, every provider, on one canvas. Local, cloud, or mixed.

Weavy — hosted SaaS if you want a managed product with credits and don’t need BYOK, local execution, or open source.

ComfyUI — deep diffusion control: samplers, VAE, ControlNet, latents.

Next steps

Edit this page on GitHub