这款免费TokenPlan里竟然有GLM5.2The free TokenPlan actually includes GLM5.2!


AI free token/AI免费积分

AI 工具 · 白嫖指南   2026 / 07  ·  阅读约 4 分钟

我使劲揉了揉眼睛。

没看错。在这个免费Token Plan的模型列表里,我拉取到了 GLM-5.2。

它安安静静地躺在免费Token Plan的可用模型列表里,官网一个字都没提。

图片

 

事情是这样的。

最近日常工作都在用主力模型,钱包确实有点遭不住了。看着账单蹭蹭往上走。

于是最近开始琢磨:能不能找一个偏向个人办公方向的免费模型,给我打辅助——日常的文档生成、报告总结这些,这些场景没必要全都烧付费 token。

目标很快锁定了商汤科技的日日新大模型的免费Token Plan,因为他们的模型是针对日常办公场景的。

ONE

配置过程:免费 Token plan 接入 Claude Code

于是我就按照官方文档,通过 CC-Switch 把日日新接到 Claude Code 上用。整体流程不复杂,十来分钟就能搞定。

一、获取 API Key

先去日日新控制台创建一个 API Key,后面备用。

地址:https://platform.sensenova.cn/console

注册登录之后,进 API Key 管理页面,创建一个新的 Key,复制下来保存好。

图片

二、通过 CC-Switch 配置日日新模型

1. 创建供应商

打开 CC-Switch,顶部工具栏选 Claude。我这边没有找到 SenseNova 的预设供应商——可能是因为不同版本 CC-Switch 的差异,或者是日日新刚公测还没来得及适配。没关系,选自定义供应商(Custom Provider),手动配置就行。

图片

 

滚动到下方,粘贴刚才申请好的 API Key。

填写 Base URL:

https://token.sensenova.cn

高级选项里的 API 格式和认证字段,用默认的就好,不用动。

图片

 

2. 拉取模型——彩蛋来了

配置好之后,拉取可用模型列表。

然后我就看到了——

图片

 

彩蛋

GLM-5.2

我使劲揉了揉眼睛。没看错,是真的 GLM-5.2。

这里多说一句:在官网上,Token plan的介绍页面并没有提到 GLM-5.2。

 

常规理解里,免费 plan 能给个基础模型就算厚道了。结果它直接塞了一个 GLM-5.2 进来——这哥们在 FrontierSWE 上超 GPT-5.5,在 SWE-Marathon 上接近 Claude Opus 4.8,智谱把它定位为目前最强的开源代码模型。

关键是:官网没说,藏了个惊喜。

那还说啥了,GLM-5.2 安排。

把 Sonnet、Opus、Haiku 的模型映射全都指向 GLM-5.2,保存配置,启用路由,搞定。

TWO

实测:让 GLM-5.2 诊断我的小程序

接入之后第一件事——拉出来遛遛。

我找了一个自己的小程序项目「咱们的账本」,把代码丢给它,让它做一次完整的项目诊断,并把结果输出成 Markdown 报告。

多说一句,「咱们的账本」是解决家庭多人共享记账的小工具——同一个账本,全家人可以一起记账、管理家庭开支。目前线上平稳运行,用户粘性也挺高,有家庭记账需求的朋友可以点链接体验一下。

回到正文,大概等了 10 分钟,报告出来了。(感觉上是有一点慢,因为项目并不大,才几十个文件。但是一想是免费的,我忍了)

报告很详尽。不是那种"这段代码写得不错,建议优化一下变量命名"的水货诊断——它真的读完了项目结构,按照严重程度分类总结,最后给出完成修复建议及修复优先级。后面准备把让它把问题都修复了。

图片

 

我之前让其他几款国产模型也跑过同一个项目的诊断。说实话,GLM-5.2 这次在诊断的详尽程度和完整度上,确实更胜一筹。

THREE

适合什么人用

想要尝鲜 GLM-5.2 但是不想直接付费买单的,可以从这里试一下。不过官网上没看到具体的额度说明,推测应该不会太高。类比可使用的deepseek,应该不会超过500次/5h,如果有测到具体额度的朋友欢迎评论区留言。

图片

 

整个配置过程十来分钟,零成本接入。免费的 GLM-5.2 接入 Claude Code 的工作流,这波怎么说都不亏。趁现在公测,先薅一波。

觉得有用?

转发给同样在烧 token 的朋友

关注「AI数字生命体」

 

AI Tools · Freebie Guide 2026 / 07 · Read in 4 minutes

I rubbed my eyes.

Nope, not misreading it. In this list of models available under the Free Token Plan, I'm seeing GLM-5.2.

It's sitting quietly in the available models list for the free token plan, with not a single mention of it on the official website.

图片

Here's what's going on.

Recently, I've been using my main model heavily for work, and the wallet is starting to feel the pinch. Watching the bill climb isn't fun.

So I started thinking: is there a free model focused on personal office use that I could use as a sidekick — for things like generating docs, summarizing reports, etc., where I don't need to burn paid tokens?

I quickly settled on SenseNova's Free Token Plan, since their models are tailored for daily office scenarios.

ONE

How to Hook It Up: Free Token Plan → Claude Code

I followed the official docs to connect SenseNova to Claude Code using CC-Switch. The process was straightforward — took maybe 10–15 minutes.

Step 1: Get an API Key

Go to the SenseNova console and create an API key — you'll need this later.

图片

Step 2: Configure via CC-Switch

图片

图片

Scroll down, paste your API key, and enter the Base URL:

https://token.sensenova.cn

图片

图片

Step 3: Fetch Models — The Good Stuff

After saving the config and fetching available models, here's where the magic happens — I see GLM-5.2.

图片

图片

And it's free? Yep, hidden gem alert.

So I mapped Sonnet/Opus/Haiku all to GLM-5.2, enabled routing, and done.

TWO

图片

First thing to do: run a diagnostic on my app "We Share Our Bills" — a family expense-sharing app that's been running smoothly.

Threw the code at GLM-5.2 and asked for a full Markdown report.

10 minutes later, the report pops out. It's *thorough* — not just "good job, clean up vars". It breaks down issues by severity, identifies root causes, and suggests fixes with priorities. This is legit diagnostics.

图片

I've tried other Chinese models before, and GLM-5.2 stands out here — much more complete and detailed than others I tested.

THREE

Who Should Use This?

If you want to try GLM-5.2 without paying, this is a great entry point. The free token limit isn't stated clearly, but based on similar services (like deepseek's 500/5h), it's probably limited — so don't go spamming it.

But for testing, for office-level code review, or for small projects? It's perfect.

Setup only took a few minutes. Free GLM-5.2 + Claude Code workflow = total win.

Go check it out while it's still in beta!

Share this with other token-burners, and follow me for more AI hacks!

As a quick side note, "Our Ledger" is a small tool for managing shared, multi-person household finances — the whole family can record and manage expenses in the same ledger. It's currently running stably online with high user retention, so families with budgeting needs can try it out via the link.

Back to the main point: I waited about 10 minutes for the report. (Felt a bit slow since the project is small — only ~50 files. But it's free, so I can’t complain.)

The report is very detailed. Not just generic advice like "improve variable naming" — it actually analyzed the project structure, classified issues by severity, and provided concrete fix recommendations with priorities. I’m planning to run the fixes next.

图片

I’ve also tested other Chinese models on the same project. Honestly, GLM-5.2 outperforms them in diagnostic depth and completeness this time.

THREE

适合什么人用

If you want to try GLM-5.2 without paying upfront, you can sign up here. No explicit quota listed, but guessing it’s limited (likely <500/5h). If anyone knows the exact limit, please share in the comments.

图片

Setup took ~10 minutes, zero cost. Free GLM-5.2 integration with Claude Code workflow — definitely worth it. Try it while it’s still in beta.

觉得有用?

Sharing with fellow token-burning devs!

关注「AI数字生命体」

暂无评论,快来发表第一条评论吧!No comments yet. Be the first!

📮 需求咨询📮 Inquiry