Who does it serve and what problem does it solve?
Chinese-language service, pay in RMB, many models on one key; subscriptions save up to 15% vs pay-as-you-go, with 1-on-1 support
Target Audience
Chinese-speaking developers and small teams using Claude Code, Codex and other AI coding tools, especially those overseas or needing stable access
How does it charge?
mixedPay-as-you-go prepaid; Pro ¥399/4 weeks (includes $440 of quota); Max ¥899/4 weeks (includes $1,040); Ultra ¥1,799/4 weeks (includes $2,120); enterprise custom
The above describes historical pricing. Check official website for current prices.
Where do users come from?
Direct 80.6%, referrals 14.5% (github.com 4.1%), brand keywords 80.6%, essentially no non-brand search; search terms include codex api, opencode api key. Social is only X at 2.9%, plus a Telegram user group. New users mostly arrive directly after word of mouth or seeing it in GitHub projects; returning users renew every 4 weeks and go straight to the console.
Channel share, initial acquisition, and growth drivers are distinct; correlation does not imply proven causation.
What remains to be verified?
The following are research questions, not established market opportunities.
- What do target users actually pay for? Is there direct user feedback or paid interview evidence?
- Are these historical prices still active today? What are the conversion, retention, and fulfillment costs?
- Is the "Other" channel still viable? What resources do indie devs need to test it?
- Whose needs are competitors failing to meet? Find specific evidence before deciding on your angle.