Midjourney
Independent research lab focused on text-to-image generation, currently on model version V7. Uses diffusion-based architecture with proprietary training. Expanding beyond stills into video generation and eventually 3D and real-time simulation. Subscription-funded, no VC backing. Known for high aesthetic quality and prompt fidelity at inference.
