PackyCode does not publish per-million-token rates, so you cannot calculate exact cost against OpenAI’s or Anthropic’s direct API pricing. The only hard figure in the data is $0/month for the free trial tier.
Free Trial: No-Card, No-Commitment
The free trial is activated without requiring a credit card. That alone puts PackyCode ahead of many relay stations that demand a top-up before you can test a single request. You authenticate, hit the endpoint, and get real GPT-4o or Claude 3.5 Sonnet responses at 172ms latency with 98% uptime. No hidden per-request overhead fees during the trial. Bottom line: Free tier exists, no card needed, no minimum recharge enforced.
Paid Tiers: Where the Costs Hide
The platform data flags “Premium pricing on MAX groups” as a con. That means the free tier gives you a taste, but sustained heavy usage — especially through the MAX quality groups — will cost more than competitors like OpenRouter or AiHubMix. What the data does not specify: batch discount thresholds or CNY minimum top-up amounts. If you plan to run 100+ concurrent Claude Code sessions, you will need to contact support or join the active community to negotiate volume pricing.
| Dimension | PackyCode | OpenRouter |
|---|---|---|
| Free tier | Yes, no card | $1 credit, card required |
| Per-million rates | Not published | Published per model |
| Batch discount | Not specified | None |
| CNY minimum top-up | Not specified | USD only |
| Hidden per-request fees | None during trial | None |
Payment Methods & CNY Constraints
You can pay with Alipay or WeChat Pay — both are standard for Chinese developers. But the data does not specify a minimum recharge amount in CNY. If you are a small team, you may end up overpaying because the lowest top-up bucket is unknown. The refund policy is also unspecified, so assume no refunds after purchase.
Pros & Cons
Pros
- Best latency for Claude Code from China: 172ms average
- Active community and fast support
- Free trial with no card required
- Supports 650+ model variants (though only GPT-4o and Claude 3.5 Sonnet are listed as primary models) Cons
- Premium pricing on MAX groups
- Some groups may switch channels without notice
- Kiro reverse-proxy groups have inconsistent quality
- Per-million rates, batch discounts, and CNY minimum recharge are not disclosed
Verdict
PackyCode is the fastest relay for Claude Code users in China, but the pricing opacity is a real problem. If you need predictable costs for production workloads, the lack of published per-token rates and batch discount thresholds means you cannot budget accurately. Use the free trial to confirm the 172ms latency and 98% uptime. Then ask the community about volume pricing before committing real CNY.
FAQ
Q: Does the free trial require a credit card? A: No. The data confirms the free trial is active without any card or top-up required. Q: Are per-million-token rates published anywhere? A: No. The platform data does not include per-token pricing for GPT-4o or Claude 3.5 Sonnet. Q: Can I pay in CNY with a minimum top-up? A: Yes, Alipay and WeChat Pay are supported, but the minimum recharge amount in CNY is not specified in the data. Q: What happens if a group switches channels without notice? A: The data lists “channel switching without notice” as a con. It means your request may route through a different backend, potentially changing latency or model behavior. Q: Is there a refund policy? A: Not specified in the data. Assume no refunds after purchase.
Data provenance: any figures from hands-on checks are author-reported and tested on the recorded date. Independent verification is unavailable unless a source is linked.
