academa-labs/manimgx is a 3D animation engine for agents that supports the Python API of Manim CE. Its rendering core is implemented in Rust and wgpu. The repository's latest release is v0.1.1, but the provided information does not describe what changed in that release. The project is written in Python and uses the MIT license.
Its goal is to make mathematical animations by reusing existing Manim code while offloading rendering to the GPU core. The README reports rendering five scenes at 1080p60, 36.2 seconds of output in total, in 3.08 seconds on an Apple M4 Pro. Under the same conditions, total render times are listed as 35.28 seconds for Manim GL, 57.61 seconds for Blender Workbench, 278.83 seconds for Manim CE, and 601.40 seconds for Blender EEVEE. Based on these results, the README compares it as 90.6x faster than Manim CE, 11.5x faster than Manim GL, and 18.7x faster than Blender Workbench. The measured scenes and settings are published in the repository's benchmarks folder.
That said, the comparison figures are the repository's own benchmarks run on the same hardware, and no independent verification is provided. The published measurement conditions are therefore a basis for considering reproduction, but not a basis for applying the performance lead to general environments.