Unverified50% confidenceFactExact time
在3D场景生成测试中，GPT-5.2 Codex出现电视机嵌入地板等空间位置错误，而Opus 4.5整体氛围感更强、3D场景还原度更高
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026
Sources
GPT-5.2 Codex实测对比Opus 4.5：编程能力、速度与体验全面评测
bilibili晟6668886/6/2026
Related Claims
UnverifiedGPT-4V 等多模态模型在直接输出屏幕坐标时存在显著的空间定位偏差75% similarUnverified在3D场景生成测试中，GPT 5.5 初版出现飞机机翼同侧放置、倒退飞行、穿模和画面卡顿等问题72% similarUnverified在3D场景迭代修正中，DeepSeek V4 一轮修正就解决了问题，而 GPT 5.5 需要多轮修改71% similarUnverifiedThe article compares GPT-5.2 Codex against Anthropic's Opus 4.5 through multiple real-world test cases including frontend generation, physics simulation, 3D scene rendering, and code refactoring.70% similarUnverifiedGPT-5.5不仅完成了同样的游戏重写任务，还额外将游戏改成了3D版本65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/52612API
curl https://kongchang.com/api/v1/knowledge/claims/52612MCP
get_claim(id=52612)