The most recent examples of the state of the art are from the initial release of GPT Astra and later Opus 5.5, maybe one or two months ago. Blender has already had an MCP since early this year, but it wasn't until these models came out that they are now capable of spatial reasoning and spatial consistency to be able to create something in 3D.
The way it works is the same way all other agentic AI works - feedback loops, and an AI driving a computer. The AI creates a shape in Blender, then takes orthographic view screenshots so it can "see" what it did, double checks its work against what the goal is, fixes errors if there are any, and then repeats until done.
It is worth noting that using Blender creates mesh representation 3D models, and does not create the boundary representation 3D models used in CAD programs. But I don't see anything stopping the industry from creating a similar MCP or other AI-native interface for b-rep CAD programs.
You can find the latest examples on social media or Youtube if you search for keywords Astra, Opus 5.5, and Blender.
This is a notably different approach than many of the existing "text to 3D" or "image to 3D" generators, which train their own machine learning algorithm to output models. Those have also been getting better, but the latest in this field was Astra driving Blender.