Back Back
Models6/day by emailReported

Video generation models can now produce visually convincing footage, but Stanford's 2026 AI Index finds they still struggle to get the physics and consistency right. On VBench-2.0, which scores videos for human-aligned faithfulness to reality rather than visual polish alone, no model evaluated in early 2026 topped a total score of Veo 3 leads the VBench-2.0 video-realism benchmark at 66.7% — no AI video model tested tops 67%

VBench-2.0 is a benchmark that evaluates video generation models on their adherence to reality, scoring them across five dimensions including Human Fidelity, Creativity, Controllability, Physics, and Commonsense. The Stanford AI Index report for 2026 states that no model evaluated in early 2026 achieved a total score above 67% on this benchmark. Google DeepMind's Veo 3 led with a score of 66.7%.

Verified 2026-09-06

Track with Insider

Why it matters

AI video can look real yet still break the laws of physics, so polish is not the same as accuracy.

FREE · NO CARD

Facts like this, six days a week

The Daily Receipts on weekdays, The Weekly Brief on Saturday — six sourced facts a day by email, two more than the free site shows. Every one traced to the filing it came from. Free.

Get the free daily brief

Free · No card · Unsubscribe anytime

Join now and you start with 100 Receipts on The Receipts Ledger — your first step toward a free year of Insider or Pro. How it works →

See today's issue →