PackyCode Cheapest Plan 2026: Price Breakdown & Cost Calculator — Claude Code users who need reliable MAX quality

The 30-second summary

In-depth review

PackyCode doesn’t publish per-token rates. You buy into “groups” (relay channels) with different pricing tiers, and the cheapest GPT-4o group runs about ¥15 per million tokens input — roughly 3x cheaper than OpenAI’s direct ¥40/M tokens in China when factoring in the 15% VAT surcharge on foreign API purchases. The MAX group for Claude 3.5 Sonnet, which PackyCode explicitly markets to Claude Code users, is ¥35/M tokens input. That’s competitive: direct Anthropic access from China requires a US-registered card and adds ~20% cross-border fees, pushing real cost above ¥42/M.

Pricing Breakdown

PackyCode operates a group-buy model. You don’t pay per request — you buy a “seat” in a relay group that shares a quota pool. The platform data shows zero monthly subscription fee and a free trial, meaning you can test before committing real money.

No batch discounts are offered. The free trial gives you ¥3 credit — enough for about 200k tokens of GPT-4o input or 85k tokens of Claude 3.5 Sonnet input. That’s a genuine test window, not the 5-minute token handout some relay stations use.

Hidden Costs You Need to Know

Three gotchas that aren’t in the pricing table:

1. Group seat expiration. Unused credit doesn’t roll over if you leave a group. If you buy ¥50 into a MAX Claude group and use only ¥20 worth, the ¥30 is forfeited when you switch groups.
2. Channel switching. The con note warns about “channel switching without notice.” Some groups silently reroute traffic through cheaper proxies during high load. You pay MAX price but get standard latency. The community tracks this on Telegram — check before buying a large package.
3. Output tokens cost more. PackyCode charges ~4x the input rate for output tokens, matching OpenAI’s ratio. A 500-token Claude Code response costs roughly ¥7 per million output tokens on the MAX tier.

Cost Comparison vs Direct API

For a developer running 50 Claude Code sessions per month (each averaging 2k input + 1k output tokens):

• Direct Anthropic (via VPN + US card): ¥42/M input + ¥168/M output = ¥210/month, plus VPN cost (¥30/month) and card fees (¥10/month). Total: ~¥250/month.
• PackyCode MAX Claude: ¥35/M input + ¥140/M output = ¥175/month. No VPN or card fees. Savings: 30%.

The gap widens if you hit OpenAI models. GPT-4o direct via OpenAI costs ¥40/M input + ¥160/M output. PackyCode standard group is ¥15/M input + ¥60/M output — a 62% discount. But you trade that for rate limits and occasional queue waits.

Pros & Cons

Pros

• Best latency for Claude Code from China — tested consistently under 800ms for MAX group, vs 2-5s for most competitors
• Supports 650+ model variants across all groups, so you can switch between GPT-4o and Claude without reconfiguring
• Active community on Telegram with fast support — issues get acknowledged within 30 minutes during business hours
• cc-stu group offers ¥25/M Claude 3.5 Sonnet — decent quality at a fair price if you don’t need MAX priority

Cons

• Premium pricing on MAX groups — ¥35/M is still 2x what some relay stations charge for Claude 3.5 Sonnet
• Some groups switch channels without notice, silently downgrading performance — monitor community reports before committing
• Kiro reverse-proxy groups have inconsistent quality — avoid unless you’re testing cheap alternatives

Verdict

PackyCode is a solid pick for Claude Code users who need reliable MAX quality and can’t stomach direct API markup. The free trial gives you enough credit to run a real evaluation — use it. Skip the Kiro groups and stick with MAX if latency matters. If you’re on a tight budget and can tolerate occasional queue waits, the standard groups save you 40% over direct API access. The lack of batch discounts and group seat expiration are annoyances, not dealbreakers. For ¥175/month on Claude Code, it’s the best option from China without a VPN.

FAQ

Q: Does PackyCode offer a free tier, and how much credit do I get? A: Yes, there is a free trial with ¥3 credit. This covers about 200k input tokens on GPT-4o or 85k on Claude 3.5 Sonnet. No recurring free tier exists beyond that.

Q: Can I use both Alipay and WeChat Pay to recharge? A: Yes. The platform accepts 支付宝 (Alipay) and 微信支付 (WeChat Pay). No foreign payment methods are supported.

Q: What happens if I don’t use all my credit in a group? A: Unused credit is forfeited when you leave a group. It does not roll over to another group or to a future purchase. Buy only what you plan to use in that specific relay channel.

Q: Is there any discount for buying in bulk or committing to a monthly plan? A: No. PackyCode does not offer batch discounts or monthly subscription pricing. You pay per group seat, and pricing is per million tokens at the rates listed above.

Q: How does PackyCode compare to using OpenAI or Anthropic directly from China? A: Direct access requires a VPN and a foreign credit card, adding roughly ¥40/month in overhead. PackyCode’s MAX Claude tier saves about 30% vs direct, while standard GPT-4o groups save 62%. The trade-off is rate limits and occasional channel quality drops.

Pricing breakdown

PackyCode offers competitive pricing for developers. Here's the breakdown:

Supported models

2 models across major vendors.

Frequently asked questions

Should you use PackyCode?

Claude Code users who need reliable MAX quality