Benchmarks

NVIDIA Researchers Introduce Physis-Lang: Self-Evolving Physical Language That Lifts Cosmos 3 Past Veo 3.1 on Physics Benchmarks

Video world models can render convincing clips that still break physics. Butter spreads like paint. Balls pass through walls. A team from NVIDIA, MIT and the University of Oxford argues the fix can come from language itself, not from extra visual, latent or numerical signals. Their framework, Physis-Lang, treats physical language as a shared, optimizable representation. The same text drives data curation, model training and inference. On the public Physics-IQ Verified leaderboard snapshot da

Get smart on it

Video generation models can create realistic-looking clips that still violate the laws of physics, such as butter spreading like paint or balls passing through walls. Physis-Lang addresses this by using language-based physics reasoning rather than additional visual or numerical signals, where text descriptions explain not just what happens in a scene but why, including entities, causes, interactions, and governing principles. The framework uses a self-evolving loop where a critic model scores captions on their accuracy and completeness against human-verified physical assertions, with the instruction prompt being refined across iterations while the base captioner remains frozen. When applied to video generation, Physis-Lang achieved top performance on physics benchmarks and outperformed a competing system on most measured tests.

Explainer: How to Actually Read an AI Benchmark (Without Being Fooled)

Read next
The biggest AI stories, once a day

One short email a day with the stories that matter. Free. Unsubscribe with one click.

Relevant courses

Want to go deeper than the news? Explore live, cohort-based AI courses taught by practitioners.

Browse AI courses on Maven
Related news
BenchmarksOpen story →

MLCommons Releases MLPerf Training v6.1 Results, Spotlighting Recent Advances in AI Training

Today, MLCommons® announced new results for the MLPerf® Training v6.1 benchmark suite. The two new benchmarks added in this round, and the submissions received, demonstrate the impact of recent architectural advances in AI training systems and the increasing diversity of the hardware used to train them. “We’ve reached an exciting new threshold for the MLPerf Training benchmark suite, with a higher level of participation in this round, the representation of both established and emerging hardw

Also reported by NVIDIA Blog

BenchmarksOpen story →

Anthropic says AI is eroding Nvidia's CUDA lead, cutting an AMD chip switch from months to three days

Anthropic claims it used AI itself to breach Nvidia's CUDA moat. The company's in-house model Mythos reportedly ported Claude to AMD GPUs in just three days, with almost no human intervention. Anthropic now runs Claude on four chip families and aims to cut costs and reduce vendor dependencies. The article Anthropic says AI is eroding Nvidia's CUDA lead, cutting an AMD chip switch from months to three days appeared first on The Decoder.

BenchmarksOpen story →

Google Deepmind's AlphaProof Code verified Signal's cryptographic core in four weeks instead of years

Google Deepmind's AlphaProof Code mathematically verified Signal's cryptographic core code in four weeks, a process that typically takes years. AI agents powered by Gemini Carbon wrote more than six million lines of proof code and flagged only about one percent for human review. Deepmind plans to extend the method to critical infrastructure like power grids and medical devices. The article Google Deepmind's AlphaProof Code verified Signal's cryptographic core in four weeks ins

BenchmarksOpen story →

A step toward defining the Standard Model on a lattice

Explore a candidate nonperturbative lattice definition of the Standard Model and how GPT-6 Astra helped turn the idea into a detailed proof.

BenchmarksOpen story →

Best AI Video Generation Models in 2026: Elo, Price per Second, Open Weights

The AI video leaderboard looks very different from 12 months ago. Native audio is now table stakes. Clips run 15–30 seconds. And an open-weights model sits in the top 3. This ranking (as of Nov 13, 2026) uses one quality signal: blind human-preference Elo from the Artificial Analysis Video Arena. We pair it with API price per second and the license status of the weights. You can check the deployment options and specs here through our GPULeaderboard.com chrome extension. Methodology Qua

BenchmarksOpen story →

XYBench: Helping with the problem behind the question

Understanding what a question actually asks matters more than raw performance scores for building reliable AI systems.