Physics in AI video: best model clears 44% of clips

A paper out on 10 November 2026 splits every scene description into physical points and checks each one with a timestamp. The top model passes 44.2% of clips, the best open model 14.9%. Here are the numbers and what they change for anyone buying realistic video.

A video frame with a tipped glass and falling ball beside a checklist where one item is crossed out
01

What researchers in Chicago measured

PhyRubric: Rubric-Grounded Diagnostic Evaluation of Physical Commonsense in Text-to-Video Generation landed on arXiv on 10 November 2026. The authors are at the University of Illinois Chicago, and they asked an awkward question: how do you tell that a model broke physics when all you have is your own eye on the clip.

The bench holds 245 descriptions, 34 categories, five areas of physics (mechanics, optics, heat, electricity, acoustics) and 2,433 generated clips. A separate training set of 3,788 videos produced PhyRubric-Judge, an open 8B model built from Qwen3-VL-8B-Instruct. On agreement with human raters it sits level with the closed Gemini 4 Argon, so a cheap open judge is enough for this kind of check.

02

Why one score out of five explains nothing

The old routine: a person watches a clip and gives it a mark. The mark is subjective, a second viewing gives another number, and it never says which principle broke. A clip where a dropped glass bounces like a ball gets a three. Whether that three is about the collision, the weight or the arc stays a guess.

PhyRubric works before anyone watches anything. Each description is split into points, some about content (what was asked for) and some about physics (what has to hold while it happens). Then the video is checked against every point separately, with a timestamp on the failure. What comes out is a diagnosis you can act on with the next prompt.

Diagram: one scene description splits into three rubric checks, one failing at a timestamp on a time ruler
03

How many models keep the story and the physics

Ten models were tested. Credit only counts when a clip matches the description and stays physically plausible. Seedance 2.5 lands 44.2%, Veo 3.1 36.2%, the best open model HunyuanVideo-1.5-8B 14.9%.

The sharper number sits next to it. Runs where the requested event never appears at all run from 20.0% to 44.5% on open models and from 5.7% to 8.6% on closed ones. That is a different failure from clumsy staging: the requested event is simply not in the frame.

04

Errors land at two poles

40.8% of the scored videos get 1 or 2 out of 5 on physics, which means a gross violation. 52.9% get a clean five. The middle is rare, and for production that is convenient: a clip either passes or it is badly broken and needs a fresh run.

05

What breaks most often

The top failure is the missing consequence at 24.0% of all failures. You ask for water being poured and the water never leaves the jug, the level in the glass never rises. You ask for a drop and the object hangs in the air. The shot exists, the cause and effect does not.

Second comes object permanence and counting at 20.4%. Things vanish between frames, double up, two glasses melt into one. In the authors' table Seedance 2.5 misses a consequence in 11.3% of videos and Veo 3.1 in 14.7%. On permanence and counting the gap is even wider: 5.2% for Seedance 2.5 against 12.9% for Veo 3.1. So it pays to compare models on the type of error that matters for your own scene.

06

What this means for buyers of realistic video

Right now physics in AI video comes down to consequence: pour, spill, drop, fall. The authors close with physical commonsense remains far from solved, and that is a fair summary, since the leader fails more than half the time.

The practice I build into a quote. Scenes with cause and effect (pour, smash, drop, open) get doubles, because the consequence often goes missing on the first run. The number of objects in frame gets counted by eye on every run, since two glasses where one was ordered is a defect viewers catch while the author does not. One good run proves nothing and needs a repeat.

When a brief says realistic video, the client usually means light, skin and texture. Physics shows up later, at approval, when the glass does not fall. That is why I show a test of the hardest scene before the main work starts.

AI video production follows the same rule: a scene is done when it repeats. For ad work with causality (a product hitting water, cream leaving a tube, a lid opening) it is cheaper to budget extra runs up front. Where physics matters and runs are tight, handcrafted AI videos take over, because there the consequence is staged by a person and shot close.

One habit that protects a reputation: quality check before publish. We already have a separate guide on that, and it pairs well with this paper. If a scene is written as code, the breakdown on a video coder is worth reading too, since some of these errors are visible before the render.

07

My take

PhyRubric earns its place because it kills the taste argument. Instead of someone saying the water looks fake, there is a point with a timestamp.

My bet for the next year: closed models still have not solved physics, so I plan scenes from simple to complex. Static shot with one object first, then motion, then two objects interacting. Interaction scenes get extra runs and a checklist at approval, with nobody grading from memory.

One more thought on client expectations. The line physically plausible gets checked on the finished file, so the person who wrote the script should be the one checking it. Showing a test of the hard scene before the main work costs less than explaining, on a locked edit, why the water is standing still.

frequently asked

01What is PhyRubric
A benchmark and a scoring method for physics in generated video. Each description is split into points in advance, and the clip is checked against them with a timestamp on every failure. It came out on 10 November 2026 from the University of Illinois Chicago.
02How bad are video models at physics
The best score is 44.2% for Seedance 2.5, then 36.2% for Veo 3.1 and 14.9% for the best open model. The authors themselves write that physical commonsense remains far from solved.
03Why are open models behind
The gap on this test is wide. Open models never show the requested event in 20.0% to 44.5% of runs, while closed models do that in 5.7% to 8.6%. The judge, though, can be open: PhyRubric-Judge at 8B matches human raters about as well as the closed Gemini 4 Argon.
04How do I check my own clip for physics
Watch the consequence and count the objects. Water poured, level rises. Object dropped, object falls. As many objects in frame as the brief asked for. A clip that got all of it first time deserves another run, since 40.8% of scored videos get 1 or 2 out of 5, which makes one clean pass a lucky pass.