WeClone Files

🎉 What's Changed

WeClone 1.0.4 tells you what a build will cost and how long it will take, before you start!

Cost and time estimate

• The confirm step of the Build portrait page estimates a build's cost, time and number of requests. A trial sends about 10 real requests to the model service to measure them, and the build reuses their answers.
• Known providers such as DeepSeek are priced at their official rates, other services at the common price across providers. For a service with off-peak prices, the estimate follows the hours a build started now would run in.
• With Codex, the estimate shows tokens and time, since Codex uses your subscription quota.
• A build starts after a trial of the saved model settings. The page tells you what is still missing before you can start.
• While a build runs, the page shows what it has spent and the time and cost still ahead, and warns you when spending runs far past the estimate. The finished page shows what the build spent.

Build portrait page

• Every new build starts from its own upload. Opening the page clears chats uploaded earlier, and a successful build clears the chats it used.
• When Codex reaches its usage limit, the build stops and keeps the answers it already got. Build again once the quota is back.

Personal assistants

• The Assistants and Portrait summary pages explain how cloud assistants such as dots, Muse and Grok Bot can use your portrait: import the exported summary into their memory or custom instructions, or send it and ask them to remember it.

Logs

• In encrypted mode, build logs now show the numbers of chats, requests and tokens, still without any chat text.

⚠️ Upgrade Notes

• A build started from the web UI reads only the chats uploaded on the Build portrait page. The csv_folder in settings.jsonc is still read by build-profile on the command line.

🐛 Bug fix

• With deepseek-v4-pro, about a quarter of extraction requests were rejected and sent again because the model wrote sample IDs as #0. These answers are accepted now (#262).

Full Changelog: https://github.com/xming521/WeClone/compare/v1.0.3...v1.0.4

😊 更新内容

WeClone 1.0.4 在开始构建前就能告诉你大概要花多少钱、多长时间！

费用和用时估算

• 「构建画像」页的确认步骤会估算这次构建的费用、用时和请求次数。试算会向模型服务发送约 10 次真实请求来测量，这些回答构建时直接沿用。
• DeepSeek 等已知服务商按官方价格计算，其他服务按各家服务商的常见价格估算。有错峰价格的服务，按现在开始构建会经过的时段计算。
• 使用 Codex 时显示 token 用量和用时，因为 Codex 用的是订阅额度。
• 开始构建前需要先用保存的模型设置试算一次，页面会提示还差哪些步骤。
• 构建过程中会显示已经花了多少、还需要多长时间和多少钱，实际花费远超估算时会提醒。构建完成后显示这次一共花了多少。

构建画像页

• 每次新建构建都从重新上传开始。打开页面时会清掉之前上传的聊天记录，构建成功后也会清掉这次用过的。
• Codex 额度用完时构建会停下，已经拿到的回答都会保留，额度恢复后重新构建即可。

个人助手

• 「助手接入」和「画像总结」页说明了 dots、Muse、Grok Bot 等云端助手怎么使用画像：把导出的画像总结导入助手的记忆或自定义指令，或者直接发给它，请它记住。

日志

• 加密模式下，构建日志也会记录聊天、请求和 token 的数量，仍然不含任何聊天内容。

升级说明

• 从网页开始的构建只读取在「构建画像」页上传的聊天记录。settings.jsonc 里的 csv_folder 仍供命令行 build-profile 读取。

问题修复

• 使用 deepseek-v4-pro 时，模型把样本编号写成 #0 这样的形式，约四分之一的抽取请求因此被判为失败并重发。现在这类回答也能正常识别（#262）。