diff --git a/src/content/posts/using-chinese-llm-apis-from-malaysia.md b/src/content/posts/using-chinese-llm-apis-from-malaysia.md
new file mode 100644
index 0000000..13a024f
--- /dev/null
+++ b/src/content/posts/using-chinese-llm-apis-from-malaysia.md
@@ -0,0 +1,130 @@
+---
+title: "Using Chinese LLM APIs From Malaysia: What Actually Works"
+description: "Using Chinese LLM APIs (GLM-5.3, DeepSeek V4) from Malaysia: how I paid in RMB, handled phone verification, and which providers actually work."
+pubDate: 2026-09-11
+category: ai
+tags: [llm, api, deepseek, glm, alipay, malaysia]
+draft: false
+---
+
+Chinese frontier models — GLM-5.3, DeepSeek V4, Qwen3.8-Max, Kimi K3 — now
+score close to Claude on coding benchmarks while costing a fraction of Western
+frontier APIs. For a solo developer or a small business running AI automation,
+that price difference is the difference between "AI that pays for itself" and
+"AI that's too expensive to run."
+
+But using them from Malaysia means crossing three walls first: paying in RMB,
+passing mainland-China phone verification, and (on some platforms) real-name
+(实名) registration. This is what I actually did, what broke, and what I'd do
+differently.
+
+## Why bother?
+
+Chinese API pricing is often 3–10× cheaper than the equivalent Western model.
+As of this writing, DeepSeek's V4.1 Flash is around **US$0.15–0.30 per 1M
+input tokens and US$0.60–1.20 per 1M output tokens** (off-peak hours are half
+price; cached input is ~20× cheaper still). A US$10 top-up goes a long way —
+tens of millions of tokens for agent workloads if you route carefully.
+
+The catch: most of these platforms are built for mainland developers. That
+means Chinese payment rails and, frequently, a mainland phone number for SMS
+verification.
+
+## Wall #1 — paying in RMB
+
+I already had my own Alipay account, which is the payment method most of these
+platforms accept (Alipay top-up on the platform itself, or Alipay to pay a
+gateway). The problem is keeping RMB in it from Malaysia.
+
+I reload my Alipay RMB balance through an exchange contact I've used for many
+top-ups. It's a common gray-market route in Malaysia — do your own due
+diligence, and never send more than you're comfortable losing. Once the RMB is
+in Alipay, paying for API credits is exactly like paying for anything else in
+China.
+
+## Wall #2 — phone verification
+
+**TokenRhythm (基元律动)** — the gateway I use most — requires a mainland China
+phone number for registration. There's no way around that requirement on their
+side, and I won't recommend dodging it. If you don't have access to a mainland
+number, platforms like DeepSeek (email signup) or BytePlus ModelArk (Singapore,
+USD billing) let you skip this wall entirely — more on those below.
+
+What made TokenRhythm worth the hassle:
+
+- **One API key, OpenAI-compatible.** It aggregates GLM-5.3, Qwen3.8-Max,
+  DeepSeek V4, Kimi K2.7-Code, MiniMax, and Seed models behind a standard
+  `https://tokenrhythm.studio/v1` endpoint. Any agent that speaks OpenAI-style
+  APIs (OpenCode, Claude Code, LiteLLM, my own bots) just works.
+- **Referral credits.** Registering through an invite link gets you ¥68 (~US$10)
+  in free credit; referring another registration gets you another ¥68. The
+  reward unlocks after you install their desktop app (OpenSquilla) and make at
+  least one call from it — after that, the same API key works in any agent.
+- **The credit is real money.** ¥68 (about US$10) covered roughly a month of
+  light agent work, or about a week of heavy sessions, with DeepSeek V4 Pro
+  through the gateway. DeepSeek's pricing is genuinely very good, so the credit
+  goes further than you'd expect.
+
+## The gotcha nobody warns you about
+
+Promotional credit on these platforms is **time-limited, not a balance**.
+TokenRhythm's new-user credit expires 30 days after it's granted — my first
+batch expired on 2026-09-03 with some still unused. Don't register and hoard
+the credit for "later". Spend it, or lose it.
+
+Also worth knowing: DeepSeek is retiring its V4 Pro model on **September 14,
+2026** — after that, `deepseek-v4-pro` requests are served by V4.1 Flash and
+billed at Flash prices. That's good news for cost, just don't expect the Pro
+behaviour to stay the same (other replacements will come).
+
+## What actually works, in order of friction
+
+1. **DeepSeek directly** — email signup, no China phone, no real-name.
+   Cheapest per-token of anything I use. The official docs list the current
+   rates with a peak/off-peak split.
+2. **TokenRhythm (基元律动)** — best model menu for the money, OpenAI-compatible,
+   but you need the mainland phone number and an invite link for the ¥68 credit.
+3. **BytePlus ModelArk** — ByteDance's international platform: Singapore
+   entity, USD billing, no China phone, and data processed in Malaysia/Singapore
+   region. Slightly pricier per token, far less friction.
+4. **OpenCode Go** — US$10/month flat for a curated roster of open coding
+   models (GLM, DeepSeek, Kimi, Qwen, MiniMax) with roughly 6× the usage value.
+   Not Chinese-billing at all, which is exactly its point.
+
+## What I'd do differently
+
+- **Check the verification requirements before signing up.** Email-only
+  platforms (DeepSeek) or international platforms (BytePlus, OpenCode Go) avoid
+  the phone wall completely. Only reach for the mainland-phone platforms when
+  their model menu or price genuinely beats the alternatives.
+- **Burn promotional credit immediately.** It's a 30-day lease, not a balance.
+- **Start with a small recharge.** Test the actual model on your real workload
+  for a week before committing.
+- **Re-check prices before each top-up.** Chinese API pricing moves fast — DeepSeek
+  cut prices permanently this year, GLM runs periodic flash promos, and deals
+  appear and vanish weekly. Never assume last month's price is this month's.
+
+## The result
+
+I now run my AI automation — Telegram support bots, agentic coding sessions,
+background LLM jobs — on Chinese LLM APIs for roughly the price of a coffee per
+week of active use. The ¥68 TokenRhythm credit alone covered a month of light
+use (a week of heavy work), and a US$10 DeepSeek top-up lasts even longer at
+V4.1 Flash prices.
+
+The walls are real, but they're payment-rail and verification walls — not
+quality walls. If you're building AI features for a business that needs to
+actually make money, the Chinese frontier APIs are worth the setup.
+
+---
+
+## Want AI automation like this for your business?
+
+I build AI-powered automations, Telegram/WhatsApp support bots, websites, and
+self-hosted infrastructure for Malaysian businesses — and I set up the cheapest
+LLM routing so your running costs stay small. If that could help your business,
+I'd love to talk:
+
+- 📱 **WhatsApp:** [+60 12-797 2969](https://wa.me/60127972969)
+- 📧 **Email:** [me@hoelee.com](mailto:me@hoelee.com)
+- 🌐 **Website:** [hoelee.com](https://hoelee.com)
diff --git a/src/content/posts/zh/using-chinese-llm-apis-from-malaysia.md b/src/content/posts/zh/using-chinese-llm-apis-from-malaysia.md
new file mode 100644
index 0000000..abfb1f2
--- /dev/null
+++ b/src/content/posts/zh/using-chinese-llm-apis-from-malaysia.md
@@ -0,0 +1,113 @@
+---
+title: "从马来西亚使用中国大模型 API：真正可行的方法"
+description: "从马来西亚使用中国大模型 API（GLM-5.3、DeepSeek V4）：我如何用人民币付款、处理手机验证，以及哪些平台真正可用。"
+pubDate: 2026-09-11
+category: ai
+tags: [llm, api, deepseek, glm, alipay, malaysia]
+draft: false
+---
+
+中国的顶级模型——GLM-5.3、DeepSeek V4、Qwen3.8-Max、Kimi K3——在编程评测中
+已经接近 Claude，价格却只有西方顶级 API 的几分之一。对于独立开发者或要靠
+AI 自动化盈利的小企业来说，这个差价就是"AI 能回本"和"AI 贵到用不起"的区别。
+
+但从马来西亚使用它们，要先跨过三道墙：用人民币付款、通过中国大陆手机号
+验证、以及部分平台要求的实名注册。下面是我实际做了什么、踩过什么坑、以及
+重来一次我会怎么改。
+
+## 为什么值得折腾？
+
+中国 API 的价格通常比同等西方模型便宜 3–10 倍。截至本文写作时，DeepSeek
+V4.1 Flash 大约是**每百万输入 tokens 0.15–0.30 美元、每百万输出 tokens
+0.60–1.20 美元**（非高峰时段半价；缓存命中的输入便宜约 20 倍）。充 10 美元
+能用很久——只要合理路由，agent 工作负载能跑数千万 tokens。
+
+问题在于：这些平台大多是为主陆开发者设计的。这意味着中国的支付渠道，
+而且经常要求大陆手机号接收短信验证码。
+
+## 第一道墙——用人民币付款
+
+我本来就有自己的支付宝账号，这也是这些平台普遍接受的支付方式（直接在
+平台充值，或用支付宝付给网关）。难点在于从马来西亚给支付宝的人民币余额
+充值。
+
+我是通过长期合作的换汇联系人给支付宝充人民币的。这是马来西亚常见的灰色
+渠道——请自行判断风险，永远不要转超过你能承受损失的钱。人民币进了支付宝
+之后，买 API 额度就跟在中国买任何东西一样简单。
+
+## 第二道墙——手机验证
+
+**基元律动（TokenRhythm）**——我最常用的聚合网关——注册时必须提供
+中国大陆手机号。这是他们平台的硬性要求，我也不打算推荐任何绕过手段。
+如果你没有大陆号码，DeepSeek（邮箱注册）或 BytePlus ModelArk（新加坡、
+美元结算）可以完全跳过这道墙——下面会详述。
+
+基元律动值得麻烦的原因：
+
+- **一个 API Key，OpenAI 兼容。** 聚合了 GLM-5.3、Qwen3.8-Max、DeepSeek V4、
+  Kimi K2.7-Code、MiniMax、Seed 等模型，统一走标准
+  `https://tokenrhythm.studio/v1` 端点。任何支持 OpenAI 风格的 agent
+  （OpenCode、Claude Code、LiteLLM、我自己写的机器人）开箱即用。
+- **邀请返利。** 通过邀请链接注册送 ¥68（约 10 美元）额度；再邀请别人注册
+  又得 ¥68。奖励需要先安装他们的桌面应用（OpenSquilla）并在里面调用至少
+  一次才能解锁——之后同一个 API Key 就能用在任何 agent 上。
+- **额度是真金白银。** ¥68（约 10 美元）大概够一个月左右的轻度 agent 工作，
+  重度使用大约一周（走网关调用 DeepSeek V4 Pro）。DeepSeek 的价格真的很
+  便宜，额度比想象中耐用。
+
+## 没人提醒你的坑
+
+这些平台的赠送额度是**限时的，不是余额**。基元律动的新用户额度在发放后
+30 天过期——我第一批额度在 2026-09-03 过期时还剩一些没用完。不要注册完
+就囤着"以后再用"，要么用掉，要么浪费掉。
+
+另外值得注意：DeepSeek 将在 **2026 年 9 月 14 日** 退役 V4 Pro 模型——
+之后 `deepseek-v4-pro` 的请求会由 V4.1 Flash 提供服务并按 Flash 价格计费。
+这对成本是好消息，只是别再指望原来的 Pro 表现（后续会有其他替代）。
+
+## 按摩擦程度排序，真正可行的方案
+
+1. **DeepSeek 官方 API** —— 邮箱注册，不需要中国手机号，不需要实名。
+   我用过的所有方案里单 token 最便宜。官方文档有当前价格（含高峰/非高峰
+   分时计价）。
+2. **基元律动（TokenRhythm）** —— 同样价格下模型菜单最全，OpenAI 兼容，
+   但需要大陆手机号和邀请链接才能拿 ¥68 额度。
+3. **BytePlus ModelArk** —— 字节跳动的国际平台：新加坡主体、美元结算、
+   不需要中国手机号，数据在新加坡/马来西亚区域处理。单 token 略贵，
+   但摩擦小得多。
+4. **OpenCode Go** —— 每月 10 美元，精选开源编程模型（GLM、DeepSeek、
+   Kimi、Qwen、MiniMax），用量价值约为订阅费的 6 倍。完全不涉及人民币
+   结算——这正是它的卖点。
+
+## 重来一次我会怎么做
+
+- **注册前先查验证要求。** 只用邮箱的平台（DeepSeek）或国际平台
+  （BytePlus、OpenCode Go）能完全避开手机号这道墙。只有当它们的模型菜单
+  或价格确实优于替代方案时，才去碰要求大陆手机号的平台。
+- **赠送额度立刻用掉。** 那是 30 天的租约，不是余额。
+- **先小额充值。** 用你的真实工作负载测试一个星期，再决定投入多少。
+- **每次充值前重新查价格。** 中国 API 价格变动很快——DeepSeek 今年已永久
+  降价，GLM 不定期做 flash 促销，优惠每周都在变。永远不要假设上个月的价格
+  还是这个月的。
+
+## 结果
+
+我现在用中国大模型 API 跑我的 AI 自动化——Telegram 客服机器人、agent 编程
+会话、后台 LLM 任务——活跃使用时每周成本大约一杯咖啡。基元律动的 ¥68 额度
+就覆盖了大约一个月（轻度）或一周（重度）的真实工作，而 10 美元的 DeepSeek
+充值按 V4.1 Flash 的价格能用更久。
+
+那些墙是真实的，但它们是支付渠道和验证的墙——不是质量的墙。如果你在为一个
+真正需要盈利的业务做 AI 功能，中国顶级 API 值得这些前期投入。
+
+---
+
+## 想要这样的 AI 自动化吗？
+
+我为马来西亚企业搭建 AI 自动化、Telegram/WhatsApp 客服机器人、网站和
+自托管基础设施——并且会帮你配置最省钱的 LLM 路由，让运行成本保持很低。
+如果你觉得这对你的业务有帮助，欢迎联系我：
+
+- 📱 **WhatsApp：** [+60 12-797 2969](https://wa.me/60127972969)
+- 📧 **邮箱：** [me@hoelee.com](mailto:me@hoelee.com)
+- 🌐 **网站：** [hoelee.com](https://hoelee.com)