Overview / start here. For the full catalog of model families see Supported Models; for connecting and configuring each provider see Providers; for the desktop download panel see Models Manager.
Run models locally, through cloud APIs with your own keys, or both in the same graph.
Local vs. cloud
Local
- Data stays on your disk
- Free after download
- Works offline
- 4–20 GB per model; needs a capable GPU or Apple Silicon
Cloud (BYOK)
- No download
- Runs on the provider’s hardware
- Latest model releases
- Billed by the provider, at the provider’s price — no NodeTool markup
- Requires internet; data goes to the provider
Mixed
Pick the best provider per node:
- ASR (Whisper) — local for sensitive audio
- Image generation — Flux locally for control, FAL/KIE cloud for speed
- Document processing — local for confidential files
Cloud models
NodeTool reaches thousands of cloud models across 30+ providers, all through the same generic nodes. The catalog below is grouped by what you’re generating — text, images, video, speech and music, 3D, embeddings. Every model runs on your own API key (BYOK): you’re billed by the provider at the provider’s price, with no NodeTool markup. Add a key in Settings → Providers and the models show up in the node’s model dropdown.
Chat models are fetched live from each provider’s API, so a provider’s list always reflects its newest releases; the families below are the ones you’ll find there. Image, video, audio, and 3D models are drawn from the manifests each provider node package ships (packages/*-nodes/), so they track what NodeTool actually exposes.
Text & chat models (LLMs)
The generic nodetool.agents.Agent and chat nodes route to whichever provider owns the model you pick. Swap claude-opus for gpt-5 or gemini-3.1-pro-preview without changing the rest of the graph.
| Provider | Model families |
|---|---|
 OpenAI |
GPT-5, GPT-5 mini, GPT-5 nano, GPT-4.1, GPT-4o, o-series reasoning models |
 Anthropic |
Claude Opus 4.x, Claude Sonnet 4.x, Claude Haiku 4.x, Claude 3.5/3.7 Sonnet |
 Google Gemini |
Gemini 3.5 Flash, Gemini 3.1 Pro, Gemini 3.1 Flash-Lite, Gemini 2.5 Pro |
 xAI |
Grok 4, Grok 3, Grok Code |
| DeepSeek | DeepSeek-V3, DeepSeek-R1 (reasoning) |
| Groq | Llama, Qwen, GPT-OSS, Kimi K2 on LPU inference |
| Mistral | Mistral Large, Mistral Medium, Mistral Small, Mixtral, Codestral, Magistral |
| Cerebras | Llama, Qwen, GPT-OSS on wafer-scale inference |
| GMI Cloud | Llama, DeepSeek, Qwen (open-weight) |
| Moonshot | Kimi K2, Kimi latest |
 MiniMax |
MiniMax-Text, MiniMax M2 |
| OpenRouter | 300+ models proxied through one key (Claude, GPT, Gemini, Llama, Qwen, DeepSeek, …) |
| Together AI | Llama, Qwen, DeepSeek, Mixtral, GLM, Kimi, and more open models |
| Evolink | GPT, Claude, Gemini, DeepSeek through one gateway key |
| kie.ai | GPT-5.5, Claude Opus/Sonnet/Haiku 4.x, Gemini 3.1 Pro (chat gateway) |
| Codex (OpenAI OAuth) | GPT chat models via your ChatGPT/Codex login |
| Claude Agent SDK | Claude via your local claude CLI subscription |
| Ollama · vLLM · LM Studio · llama.cpp | Any local open-weight model — Llama, Qwen, Gemma, DeepSeek, Mistral, Phi, GPT-OSS |
| HuggingFace | Chat inference over Hub models (500,000+) |
Reasoning, tool calling, and vision input are exposed where the provider supports them. For local text generation without an API key, see Supported Models (llama.cpp, MLX, Transformers).
Image generation models
Generate and edit images through nodetool.image.TextToImage and nodetool.image.ImageToImage.
Flagship models
| Model | Provider | Capabilities | Key features |
|---|---|---|---|
| Black Forest Labs FLUX.2 |  BFL |
T2I with control | Photoreal images, multi-reference consistency, accurate text, flexible control |
| Google Nano Banana 2.0 / Pro | Google |
High-res T2I/Edit | 2K output, 4K upscaling, better text and character consistency |
| GPT Image 2 | OpenAI |
T2I/Edit | Photorealistic generation and instruction-based editing |
| Ideogram V3 | Ideogram | T2I/Edit | Typography rendering and artistic style control |
| Seedream 4.5 | ByteDance | T2I/Edit | High-fidelity generation and instruction-based editing |
| Z-Image Turbo | Z-AI | T2I | Fast generation with strong prompt adherence |
Full catalog by provider
| Provider | Image models |
|---|---|
| OpenAI |
GPT Image 2, GPT Image 1.5, GPT Image 1, GPT Image 1 mini |
| Google |
Gemini 3.1 Flash Image (Nano Banana 2), Gemini 3 Pro Image (Nano Banana Pro) |
| Black Forest Labs |
FLUX.1 [schnell], FLUX.1 [dev], FLUX.1 Pro, FLUX 1.1 Pro, FLUX 1.1 Pro Ultra, FLUX.2 [dev/flex/pro/max], FLUX.2 Klein, FLUX Kontext [pro/max], FLUX Fill, FLUX Canny, FLUX Depth, FLUX Redux, FLUX PuLID |
| Stability AI | Stable Diffusion 1.5, SD 2.1, SDXL, SDXL Lightning, Stable Diffusion 3 Medium, SD 3.5 [medium/large/large turbo], Stable Cascade |
| Qwen | Qwen Image, Qwen Image Edit (2509/2511/Plus), Qwen Image Max, Qwen Image Layered |
| ByteDance | Seedream 3.0, Seedream 4.0, Seedream 4.5, Seedream 5 Lite |
| Ideogram | Ideogram V2, V2 Turbo, V2a, V3 [balanced/quality/turbo], Ideogram Character |
| Recraft | Recraft V3, Recraft 20B, Recraft V4, Recraft V4 Pro (raster + SVG) |
| Others | Kolors, HiDream, Hunyuan Image 3, OmniGen v1/v2, Sana, PixArt-Σ, Aura Flow, CogView4, Luma Photon, GLM Image, Reve, Bria, ERNIE Image, Emu 3.5, Lumina, F-Lite, Fibo, Playground v2.5, Juggernaut, DreamShaper, Proteus, Recraft, Chroma, Janus, MiniMax Image-01, xAI Grok Imagine |
| Upscale & restore | Topaz, Real-ESRGAN, ESRGAN, SUPIR, Clarity Upscaler, SeedVR, GFPGAN, CodeFormer, AuraSR, SwinIR, DDColor |
Access aggregators — 
FAL (450+ image endpoints), Replicate (170+), and kie.ai — carry most of these families plus hundreds of community fine-tunes, LoRAs, and control variants.
Video generation models
Generate video through nodetool.video.TextToVideo and nodetool.video.ImageToVideo.
Flagship models
| Model | Provider | Capabilities | Key features |
|---|---|---|---|
| OpenAI Sora 2 Pro | OpenAI |
T2V/I2V up to 15s | Realistic motion, refined physics, synchronized native audio, 1080p |
| Google Veo 3.1 | Google |
T2V/I2V with references | Multi-image references, extended length, native 1080p with synced audio |
| ByteDance Seedance 2.0 | ByteDance | T2V/I2V | Cinematic video with stable characters and smooth motion |
| Runway Gen-4 / Aleph | Runway | T2V/I2V/Extend | Precise motion control and high fidelity |
| Luma Ray 2 | Luma AI | T2V/I2V/edit | Fast generation and creative editing |
| xAI Grok Imagine | xAI |
T2V/I2V/T2I | Coherent motion with synchronized audio |
| Alibaba Wan 2.6 |  Alibaba |
Multi-shot T2V/I2V | Affordable 1080p, stable characters, native audio |
| MiniMax Hailuo 2.3 | MiniMax |
High-fidelity T2V/I2V | Expressive characters, complex motion and lighting |
| Kling 3.0 |  Kling |
T2V/I2V with audio | Synchronized speech, ambient sound, and effects |
Full catalog by provider
| Provider | Video models |
|---|---|
| OpenAI |
Sora 2, Sora 2 Pro |
| Google |
Veo 3.1, Veo 3.1 Fast, Veo 3.1 Lite |
| Kling |
Kling 2.1 [standard/pro/master], Kling 2.5 Turbo Pro, Kling 2.6, Kling 3.0, Kling Avatar, Kling Lip Sync |
| ByteDance | Seedance 1.0 [lite/pro], Seedance 1.5 Pro, Seedance 2.0 |
| Alibaba |
Wan 2.1, Wan 2.2, Wan 2.5, Wan 2.6, Wan VACE |
| MiniMax |
Hailuo 02, Hailuo 2.3, Video-01 [director/live] |
| Runway | Gen-3 Alpha, Gen-4, Gen-4 Turbo, Gen-4.5, Aleph |
| Luma AI | Dream Machine, Ray 2 [540p/720p], Ray Flash 2 |
| PixVerse | PixVerse v4, v4.5, v5, v5.6, v6 |
| Vidu | Vidu 2.0, Vidu Q1 |
| Others | LTX Video, LTX-2, Hunyuan Video (+ v1.5, Avatar, Foley), Mochi v1, CogVideoX-5B, Marey, Lucy, Magi, Kandinsky 5, Sana Video, SkyReels, Stable Video Diffusion, Zeroscope, Pika, xAI Grok Imagine |
| Lip sync & avatars | LatentSync, Lipsync 2 [pro], MultiTalk, OmniHuman, SadTalker, Live Portrait, Stable Avatar, DreamActor, InfiniteTalk |
| Upscale | Topaz Video, SeedVR, Crystal Video Upscaler |
Speech, audio & music models
Generate speech through nodetool.audio.TextToSpeech, transcribe with nodetool.text.AutomaticSpeechRecognition, and create music through provider music nodes.
Text-to-speech (TTS)
| Provider | TTS models |
|---|---|
| OpenAI |
TTS 1, TTS 1 HD, gpt-4o-mini-tts |
 ElevenLabs |
Eleven v3, Turbo v2.5, Flash v2.5, Multilingual v2 |
| Google |
Gemini TTS, Gemini 3.1 Flash TTS |
| MiniMax |
Speech 02 [hd/turbo], Speech 2.6, Speech 2.8 |
| Open models | Kokoro, Orpheus, Chatterbox [HD/Pro/Turbo/Multilingual], Dia, F5-TTS, XTTS v2, VibeVoice, Zonos, CSM-1B, Index-TTS 2, Qwen 3 TTS, Maya, StyleTTS2, Parler-TTS, Tortoise, VoiceCraft, OpenVoice, Cartesia Sonic |
Speech recognition (ASR)
| Provider | ASR models |
|---|---|
| OpenAI |
Whisper |
| Google |
Gemini audio transcription |
| ElevenLabs |
Scribe (speech-to-text) |
| Together AI | Whisper Large v3, Voxtral, Parakeet |
| HuggingFace | Whisper and other ASR models, plus MLX Whisper locally |
Music & sound
| Provider | Music & sound models |
|---|---|
| Suno (via kie.ai) | Suno — song generation, extend, cover, remix |
| ElevenLabs |
V3 Dialogue, Sound Effects |
| MiniMax |
Music 01, Music 1.5, Music 2.6 |
| Google |
Lyria 3 Clip, Lyria 3 Pro |
| Open models | MusicGen, Stable Audio 2.5, ACE-Step, DiffRhythm, YuE, Riffusion, Flux Music, MMAudio, ThinkSound |
3D generation models
Generate 3D assets through nodetool.3d.TextTo3D and nodetool.3d.ImageTo3D.
| Model | Provider | Capabilities | Key features |
|---|---|---|---|
| Meshy AI | Meshy (MESHY_API_KEY) |
T2M/I2M | Textured mesh generation |
| Rodin AI | Rodin (RODIN_API_KEY) |
T2M/I2M | High-fidelity 3D creation |
Open 3D families — Hunyuan3D (v2.1/v3), Trellis / Trellis 2, TripoSR / Tripo, Shap-E, Point-E, OmniPart, Era3D — run through HuggingFace / base-node 3D nodes (HFTextTo3D, HFImageTo3D) and FAL/Replicate rather than dedicated runtime providers. See Providers for details.
Embedding models
Power RAG and semantic search through embedding nodes.
| Provider | Embedding models |
|---|---|
| OpenAI |
text-embedding-3-small, text-embedding-3-large, ada-002 |
| Google |
gemini-embedding-2 |
| Mistral | mistral-embed |
| Cohere | embed-v4.0, embed-english-v3.0, embed-multilingual-v3.0 |
| Voyage AI | voyage-3.5 and the Voyage line |
| Jina AI | jina-embeddings-v3 |
| Together AI | m2-bert, BGE, and other open embedding models |
| Ollama · HuggingFace | nomic-embed, mxbai-embed, sentence-transformers (local, no key) |
Using these models
Access these models through NodeTool’s generic nodes:
- For Video: Use
nodetool.video.TextToVideoornodetool.video.ImageToVideo - For Images: Use
nodetool.image.TextToImage - For 3D: Use
nodetool.3d.TextTo3Dornodetool.3d.ImageTo3D - For Music: Use kie.ai-backed Suno nodes (Suno Generate, Extend, Cover)
- Select Provider: Click the model dropdown in the node properties
- Configure API: Add provider API keys in
Settings → Providers
Access via kie.ai (recommended for broad model support): Many of these models are available through kie.ai, an AI provider aggregator that often offers competitive or lower pricing compared to upstream providers.
- Configure using
KIE_API_KEYinSettings → Providers
Access via fal.ai:
- Configure using
FAL_API_KEYinSettings → Providers
Cost Considerations: Cloud models typically charge per generation. Check each provider’s pricing before extensive use. Local models are free after download but require capable hardware.
Getting Started
Option 1: Start with Local Models (Recommended)
- Open Models → Model Manager in NodeTool
- Install these starter models:
- GPT-OSS (~4 GB) – Text generation and chat
- Flux (~12 GB) – High-quality image generation
- Wait for downloads to complete
- Run templates – they’ll work offline!
Option 2: Start with Cloud Providers
- Get an API key from a provider:
- In NodeTool, go to Settings → Providers
- Paste your API key
- Select the provider when using AI nodes
Detailed Guides
General
- Models Manager – Download and manage AI models
- Getting Started – First workflow
Local AI
- Supported Models – List of local models (llama.cpp, MLX, Whisper, Flux)
Cloud AI
- Providers Guide – Set up OpenAI, Anthropic, Google
- HuggingFace Integration – Access 500,000+ models
Advanced
- Self-Hosted Deployment – Secure deployments
- Deployment Guide – Cloud infrastructure