Tencent Hunyuan Hy3 Matches Gemini 3.5 in Local Physics Simulation
rohanpaul_ai · x · 2026-07-07
In a test conducted on atomic.chat, a desktop app for running local LLMs, multiple models were tasked with generating physics simulations for objects like bowling balls, hockey pucks, and billiard balls. Tencent's newly released Hunyuan Hy3 achieved physics quality close to Gemini 3.5 at approximately 1/35th of the cost. The difficulty of the test lies in maintaining causal physical relationships, such as collision timing, momentum transfer, friction, and plausible scattering—the billiard break shot is particularly unforgiving of any angular errors. Notably, DeepSeek-V4 consumed the most tokens (50,600).
Related event: Tencent Open-Sources Hunyuan Hy3 for Agentic Workloads(26 posts)→
More from Models
- Same Echo Maze prompt, three frontier models: all passed visually but shipped the same hidden bug — eyishazyer · 2026-09-11
- Benchmark scores drop from 89% to 19% on new evals — how benchmaxxing breaks leaderboard trust — airesearch12 · 2026-09-11
- ChatGPT tells user their question is too hard and to 'accept dumber answers' — phido3000 · 2026-09-11
- Developer Building a Unified Leaderboard of All Model Benchmark Scores — airesearch12 · 2026-09-11
- Rumor claims Kimi faked performance by serving Claude; DeepSeek new model surprises in evals — realsohamparekh · 2026-09-11
- GPT-5.6 writes well but is instantly forgettable, user complains — BasedRaddka · 2026-09-11