The results highlight a trend of European AI companies competing with U.S. and Chinese firms in non-LLM sectors. Black Forest Labs, based in Freiburg, Germany, achieved this ranking on the same day that ElevenLabs' v4 Turbo model topped a leaderboard for text-to-speech AI. While Europe lacks its own frontier large language model, these firms are targeting specialized disciplines like physically accurate world modeling and audio synthesis to maintain competitiveness.
Tuesday, October 20, 2026
Story
FLUX 3 Beats Gemini and Sora 2 on Google Physics-IQ Benchmark
AI14 days ago
A video model created by Black Forest Labs now occupies the top spot in a test gauging how well AI predicts the laws of nature. FLUX 3 has claimed the #1 position on the Physics-IQ benchmark from Google DeepMind, which evaluates the ability of models to predict the outcome of filmed experiments in fields such as thermodynamics, magnetism, and fluid dynamics. The system outperformed other frontier models, including Gemini Omni 1.1 Flash, Sora 2, Veo 3.1, MiniMax H3, Seedance 2.5, and Cosmos3, with a physics verification layer further increasing scores over the base model.
Looking for more context or analysis?
Tilt chat provides reliable, tailored insights, alerts, and automations.