Into the Flow.
Press play. Follow the spiral.
One thing to notice
The fast region shrinks.
The fluid keeps escaping.
Follow a bright trace toward the center. Watch it spiral inward, turn, and leave above or below. Then the view opens: five slices let you look inside the same flow. The region of intense motion gets smaller while fluid continues through it.
Now watch the whole thing again.
A little further in
What am I looking at?
Colored traces reveal a separately authored fluid model: an inward spiral near the center, with outflows in opposite directions along its axis. The model preserves fluid volume. The shrinking core is a region of intense motion, with fluid passing through it.
The two halves separate so you can see inside. That is an exploded view: the fluid itself is not tearing apart. The five slices show relative speed on a shared scale that changes with the model’s swirl scale. Their contour lines connect equal speeds; they are not paths through the fluid. The outer traces use color to distinguish groups of paths.
The camera moves closer to keep the shrinking region visible. The radius and swirl readouts show the model’s prescribed scales. Traces appear and fade to keep the flow readable, rather than fluid being created or destroyed. The final frame still shows a finite state.
Why Navier–Stokes?
The Navier–Stokes equations describe fluid motion. In a September 8, 2026 proposal, OpenAI presented a proof that a fluid driven by a smooth external force can reach unbounded speed in finite time, even while its total kinetic energy stays bounded.
The inward spiral and opposed outflows take inspiration from section 2.1 and Figure 1 of that proposal. This film uses a separate illustrative model. It does not render the paper’s constructed solution or validate its result.