Simpler: let the ZeroGPU allocator queue GPU calls (concurrency_limit=None), parallel CPU events

Multi-user concurrency: compose outside the GPU queue, parallel assists/examples/videos

Update app.py

Update app.py

Reorder composer providers from live measurements (#3)

Fix composer hangs and JSON failures (no timeout on the LLM call) (#2)

Update app.py

Load weights from MiniMaxAI/MiniMax-Music3 (#1)

Update README.md

Update app.py

Update app.py

Update app.py

Replace H1 title with theme-aware logo (600px, centered)

Update README.md

MiniMax Music 3 demo (modular diffusers pipeline + AoTI)

initial commit