腾讯混元Hy3以1/35成本达到Gemini 3.5物理模拟水平

Rohan Paul · @rohanpaul_ai · X·2026-07-07 04:16·57天前
AI 导读

在atomic.chat桌面应用的物理模拟测试中,腾讯混元Hy3仅用29,757 tokens(成本$0.006)即实现与Gemini 3.5(23,300 tokens,$0.21)相当的保龄球、空气曲棍球和台球HTML5 Canvas物理效果,碰撞与动量传递真实无穿模。GLM-5.2(25,454 tokens,$0.07)编码能力较强但物理表现一般;DeepSeek-V4消耗50,600 tokens($0.009)却输出最弱的物理场景。

Rohan Paul@rohanpaul_ai
51AI 编辑部评分,满分 100

腾讯混元Hy3以1/35成本达到Gemini 3.5物理模拟水平

2026-07-07 04:16· 57天前
AI 导读

在atomic.chat桌面应用的物理模拟测试中,腾讯混元Hy3仅用29,757 tokens(成本$0.006)即实现与Gemini 3.5(23,300 tokens,$0.21)相当的保龄球、空气曲棍球和台球HTML5 Canvas物理效果,碰撞与动量传递真实无穿模。GLM-5.2(25,454 tokens,$0.07)编码能力较强但物理表现一般;DeepSeek-V4消耗50,600 tokens($0.009)却输出最弱的物理场景。

More good news for local LLMs.

Tencent’s new Hunyuan Hy3 reaches Gemini 3.5-level physics quality for 35x less cost.

Test was done on atomic[.]chat, a desktop app that runs LLMs locally.

The prompt asked 4 models to build bowling, air hockey, and pool simulations. The harder part was preserving physical cause and effect.

A strike needs collision timing, mass transfer, pin rotation, friction, and believable scattering.

A pool break exposes the same weakness, because every wrong angle compounds immediately.

Interestingly, DeepSeek-V4 spent the highest number of tokens (50,600 ), yet produced the weakest visual physics in this test.

atomic.chatNew Hunyuan Hy3 hits Gemini 3.5 quality on physics for 35x cheaper! We gave 4 models the same prompt: build three self-contained HTML5 canvas scenes with real p...

来源:Rohan Paul· x.com