使用sub2api对gpt pro20x 进行的中专，生成一些流程图的时候总是超时，但是生成描述复杂的卡通图的时候不会超时，百思不得其解，麻烦各位佬试试你们会超时吗？应该怎么解决这个问题，快被折磨死了5555
提示词如下：
No marketing poster style, no exaggerated claims, no logos, no watermarks, no fake screenshots, no illegible dense text, no decorative gradients, no cartoon style. Use precise academic engineering diagram aesthetics.Use case: infographic-diagram
Asset type: data model figure
Primary request: Show content-addressed assets identified by sha256 hashes, shared cfg/csv/eva assets, workspace manifests referencing assets, multiple analysis sessions reusing the same asset.
Composition: node-link data model, cache store in center, sessions around it.
Style: clean academic data architecture, white background, thin gray connectors.
Text requirement: in-image text should be primarily Simplified Chinese; only keep technical terms such as TraceLab, PTU, UOS, Linux, Windows, React, Tauri, Rust, IPC, CFG/CSV, XML/EVA/EVB, sha256, TEG, x86_64, ARM64, LoongArch in English where appropriate.
Use these exact Chinese labels where text appears: 内容寻址资产库, sha256 标识, 共享数据资产, CFG/CSV 资产, XML/EVA/EVB 资产, workspace manifest, 分析会话 A, 分析会话 B, 引用关系, 复用资产.
Constraints: no storage saving percentage claims.
Size: 2048x1152
Negative constraints: No marketing poster style, no exaggerated claims, no logos, no watermarks, no fake screenshots, no illegible dense text, no decorative gradients, no cartoon style. Use precise academic engineering diagram aesthetics.
8 个帖子 - 3 位参与者
来源: LinuxDo 最新话题查看原文