Midjourney
Midjourney is a proprietary text-to-image AI service accessible via Discord, not a local model. Operators cannot download or run it on their own hardware; they submit prompts through a Discord bot and receive generated images. The service uses a closed-source diffusion model trained by Midjourney Inc. and offers subscription tiers for commercial use. Unlike local alternatives like Stable Diffusion, Midjourney provides no control over the underlying model, inference settings, or privacy of prompts.
Practical example
An operator wanting to generate an image of "a cat wearing a spacesuit" would join the Midjourney Discord server and type /imagine prompt: a cat wearing a spacesuit. The bot processes the prompt on Midjourney's servers and returns four image variants. The operator can upscale or vary selections, but cannot modify the model, adjust sampler settings, or run offline. This contrasts with local tools like Stable Diffusion WebUI, where the operator controls every parameter and runs on their own GPU.
Workflow example
In a typical workflow, an operator opens Discord, navigates to a Midjourney channel, and sends /imagine prompt: .... The bot replies with a progress indicator and eventually posts a 4-image grid. The operator then reacts with U1–U4 to upscale or V1–V4 to create variations. All processing happens remotely; the operator's hardware only runs Discord. For local alternatives, an operator would use Stable Diffusion WebUI or ComfyUI, loading a model like SDXL into VRAM (e.g., ~7 GB for SDXL base) and generating locally.
Reviewed by Fredoline Eruo. See our editorial policy.