Node-based AI workflow builder

Every AI model.
One canvas.

Build visual workflows that connect any AI model — from OpenAI and Claude to free local models running on your machine. Drag, connect, run. No code required.

Free · Alpha coming soon

Works with every AI provider

OpenAIAnthropicGoogleMetaMistralStability AIHugging FaceOllama

Your models. Your choice.

Pexper connects to every major AI provider — cloud APIs, ML platforms, and models running locally on your hardware. Use what you want, switch anytime.

API Providers

Connect your API key and access any model instantly.

OpenAI

GPT-4o, GPT-4.1, o3, DALL-E 3, Whisper, TTS

Text, images, audio, vision, reasoning

Anthropic

Claude Opus, Sonnet, Haiku

Text, vision, reasoning, analysis

Google

Gemini 2.5 Pro, Flash, Imagen, Veo

Text, images, video, audio, code

Mistral

Large, Medium, Small, Codestral

Text, code, multilingual

Cohere

Command R+, Embed, Rerank

Text, RAG, embeddings

AI21

Jamba 1.5

Text, long context

ML Platforms

Access thousands of open-source models through ML platforms.

Hugging Face

400K+ models — text, image, audio, video.

Replicate

Run open-source models in the cloud with a single API call.

OpenRouter

One API, hundreds of models. Automatic fallback and routing.

Together AI

Fast inference for Llama, Mistral, Stable Diffusion, and more.

Fireworks

High-speed inference optimized for production workloads.

Local / On-Device

Run models locally — 100% private, zero API costs.

Ollama

One-line model downloads. Run Llama 3, Mistral, Phi, Gemma from your terminal.

LM Studio

Desktop app for discovering, downloading, and running local LLMs.

Apple MLX

Native Apple Silicon inference — fast, efficient, optimized for M-series.

llama.cpp

Lightweight C++ inference. Maximum control, minimum overhead.

Osaurus

macOS-native AI runtime with autonomous agents and persistent memory.

Your data never leaves your machine. No API keys needed. No internet required.

Three steps. Any workflow.

1
Cards being dragged from sidebar onto canvas

Drag cards onto the canvas

Pick from generation, data, chat, agent, and analysis cards. Each card is a building block.

2
Nodes with connection lines, data flowing

Connect them into a workflow

Draw connections between cards to define data flow. Branch, merge, loop — build logic visually.

3
Run button with green play icon

Run — or save as a reusable app

Hit run and watch your workflow execute in real time. Save it as a reusable app for different inputs.

Everything you need, in five card types

Every card is a node on the canvas. Connect them in any combination to build workflows for any task.

Generation

Create text, images, video, and audio with any model. Mix and match to get exactly the output you need.

Text · Image · Video · Audio · Music

🗃

Data Storage

Load, save, transform, and organize your data. The backbone of any workflow.

Files · JSON · CSV · PDF · Database · Variables

💬

Chat

Conversations with thinking, reasoning, and memory. Chain-of-thought for complex problem-solving.

Conversation · System prompts · Thinking · Memory

🌐

Agent

Autonomous tools that search, browse, and act. Multi-step tasks completed independently.

Web search · Crawl · Code execution · Tool use

🔍

Analysis

Extract, classify, and understand content. Turn raw data into usable information.

OCR · Speech recognition · Classification · Extraction

Every format your workflow needs

TXT MD JSON CSV HTML DOCX PDF XLSX PNG JPG WEBP SVG GIF MP3 WAV FLAC MP4 WEBM MOV

See it in action

The canvas is where it all comes together. Drag cards, draw connections, see your data flow in real time.

Product screenshot or video: canvas with nodes, connections, card inspector, output preview

Built different

"AI tools charge $50+ a month."

Pexper is free. Bring your own API keys and pay providers directly — no markup. Or use local models for $0.

"I'm locked into one provider."

Pexper works with every provider. Switch from OpenAI to Claude to a local model in one click. No migration, no lock-in.

"I don't trust the cloud with my data."

Run models entirely on your hardware. Your prompts, your files, your outputs — nothing leaves your machine.

"I'm not a developer."

You don't need to be. Connect your existing ChatGPT or Claude subscription directly — no API keys, no terminal commands, no setup.

"I need text AND images AND video AND audio."

One canvas. Every format. Every model. Build workflows that mix text, images, audio, video, web search, and data analysis in a single flow.

What's new in Pexper

Feature Video generation cards with Runway and Veo support
Improvement Faster Ollama integration, 3x quicker model loading
Provider Added Mistral Large 2 and Cohere Command R+ support
Feature Workflow export — save any workflow as a reusable app

Be the first to build with Pexper.

Free alpha coming soon. Get notified when it's ready.

Join creators on the waitlist