Search

AutoGPT

Free trial

AutoGPT began as an open-source autonomous agent project and now ships a hosted platform built around named AI specialists. You pick an expert for content, sales, research or administration, brief it in plain language, and connect the accounts it may use. A coordinating assistant called Otto routes requests to the right specialist and can pull several into one job. Experts run on schedules, and each has its own spending approval threshold. A workflow builder remains available for defining steps by hand, and the whole platform can still be self-hosted with Docker.

n8n

Freemium

n8n is a workflow automation tool you can run in the cloud or host yourself. Workflows are built on a visual canvas with more than 1,000 nodes, and each run counts as one execution no matter how many steps it contains. The AI agent node, MCP server trigger and chat hosting place language models inside ordinary business automations, connected to databases, CRMs and internal APIs. Code steps support JavaScript and Python when the visual builder runs out of room. The Community Edition is free to self-host, and cloud plans are priced by monthly executions.

CrewAI

Freemium

CrewAI provides both an open-source Python framework and a hosted studio for building multi-agent systems. In the studio you describe a process in plain language or sketch it on a canvas, and the builder assembles agents with roles, tools and an order of steps. Workflows export as a Python project, an MCP server or a UI component, and connect to GitHub for version control. The runtime records every model and tool call for tracing, and adds approval gates, PII redaction and role-based access for teams that need oversight before agents reach production systems.

Zed

Freemium

Zed is an open-source code editor built for speed, with a native interface and GPU-accelerated rendering. It includes an agent panel that can read your project, edit files, and run commands, plus edit predictions that suggest the next change as you type. Zed connects to hosted models through a Pro subscription or to your own keys from Anthropic, OpenAI, Google, Bedrock, Ollama, and other providers. It also runs external agents such as Claude Agent and Codex CLI through the Agent Client Protocol. Multiplayer editing lets several developers share one workspace in real time.

FLUX

Paid

FLUX is Black Forest Labs' family of image models, reachable through an API, a browser playground or downloaded open weights. It covers text-to-image and image-to-image, with the FLUX 3 line handling prompts, reference images and outputs up to 4K. Klein versions are smaller and cheaper, and commercial licences allow fine-tuning and LoRA training on your own hardware. Developers pay per generated image rather than a subscription, with prices set by model and resolution. The playground runs in the browser, so you can test prompts before writing code; teams wanting a fixed monthly bill need to estimate volume first.

Stability AI

Freemium

Stability AI develops the Stable Diffusion family of open image models and a hosted platform around them. The web app generates images from text and images, with editing tools for inpainting, upscaling and structure control. Developers can access the same models through an API for products and pipelines, and many weights are published openly for self-hosting. The company also offers video and audio models. Paid credits cover hosted generation; the open models are free to run on your own hardware under their licences. It suits teams that want flexibility between a managed service and running models themselves, from indie developers to studios with GPU infrastructure.

Ollama

Free

Ollama lets you run open-weight language models on your own machine with a single command. It handles model downloads, GPU acceleration and local serving, exposing an API compatible with common OpenAI-style clients. The library includes models for chat, coding, vision and embeddings, from laptop-friendly options to larger models for workstations. Because everything runs locally, prompts and data never leave your computer — useful for confidential work, offline use and experimentation without per-token costs. It runs on macOS, Linux and Windows, and integrates with tools like Open WebUI. It is the simplest way to start with local LLMs, though large models need substantial RAM and a capable GPU.