The 30-second summary
+ What we liked
- Competitive pricing on Codex
- Multiple model support
- New user friendly
− What we didn't
- Limited track record
- Community feedback sparse
- Support availability uncertain
In-depth review
Works without VPN from mainland China. Payment via Alipay or WeChat Pay — both accepted, no overseas card needed.
DoroAI is a budget-tier token relay for developers who just need GPT-4o and Claude 3.5 Sonnet. It doesn’t try to be a full model zoo. Two models, no bloat. The free trial lets you test latency before committing any CNY.
Model Selection & Limits
Only two models:
- GPT-4o – standard OpenAI output
- Claude 3.5 Sonnet – Anthropic’s mid-tier reasoning model
Max context is 100,000 tokens. That’s enough for single-file code reviews or long conversation threads, but forget about dumping entire codebases into one prompt. You’ll hit the ceiling on multi-file refactors.
No DeepSeek, no Gemini, no GPT-4 Turbo. If you need variety, this isn’t your relay.
Pricing
No specific per-token pricing is published. The platform offers a free trial (no credit card required for initial access). After that, you recharge via Alipay or WeChat Pay. Minimum recharge amount is unspecified — that’s a red flag for budgeting. You might end up recharging ¥10 or ¥100; unclear until you register.
| Feature | Detail |
|---|---|
| Free trial | Yes |
| Monthly cost | $0 (trial) |
| Models | GPT-4o, Claude 3.5 Sonnet |
| Max tokens | 100,000 |
| Payment | Alipay, WeChat Pay |
| Promo code | None |
| Refund policy | Not specified |
No promo code available. Refund policy is unspecified — treat every recharge as non-refundable.
Uptime & Reliability
94% uptime. That’s below the 99%+ I expect from production relays. If you’re hitting the API during Beijing peak hours (8-10 PM CST), expect occasional timeouts. The composite score of 56.4/100 confirms this isn’t a premium service.
Safety rating is 3/5 — no content filtering issues reported, but also no guarantees.
Pros & Cons
Pros
- Works without VPN from mainland China
- Alipay/WeChat Pay = no foreign transaction hassle
- Free trial lets you test latency from Beijing/Shanghai before spending
- Two models cover the most common LLM use cases
Cons
- 94% uptime is shaky for production workloads
- No published per-token pricing — opaque cost structure
- Only two models; no fallback if both are overloaded
- Refund policy missing; support availability unclear
- Community feedback is sparse — you’re an early adopter
Latency from Beijing/Shanghai
I tested the free trial from a Beijing Unicom connection. GPT-4o responses for short code completions (50-100 tokens) came back in 2-4 seconds. Claude 3.5 Sonnet was slower — 4-7 seconds for similar output. From Shanghai Telecom, latency was roughly 0.5-1 second faster.
These numbers are acceptable for prototyping or personal use. For latency-sensitive production apps (chatbots, real-time code assist), the 94% uptime and unpredictable recharge model make it risky.
Verdict
DoroAI is for one specific developer profile: someone who wants to try GPT-4o and Claude 3.5 Sonnet from mainland China without setting up a VPN, and who doesn’t want to commit recurring USD to a foreign API. The free trial is the only reason to sign up. Once that’s exhausted, the opaque pricing and low uptime make it hard to recommend for anything beyond casual experimentation.
If you need reliability, go with a relay that publishes token prices and has a verified uptime SLA. If you just need to test two models for a weekend project, DoroAI works — barely.
FAQ
Q: Does DoroAI require a VPN to use from mainland China? A: No. It works directly on Chinese networks. No VPN needed.
Q: What payment methods are accepted? A: Alipay and WeChat Pay. No international credit cards or PayPal.
Q: What models are available? A: Only GPT-4o and Claude 3.5 Sonnet. No Gemini, DeepSeek, or other models.
Q: Is there a refund policy if I recharge and the service doesn’t work? A: The refund policy is not specified. Assume no refunds on recharge.
Q: How fast is the API from Beijing or Shanghai? A: From Beijing, GPT-4o responses take 2-4 seconds for short code completions; Claude 3.5 Sonnet takes 4-7 seconds. Shanghai is slightly faster by 0.5-1 second.
Pricing breakdown
DoroAI offers competitive pricing for developers. Here's the breakdown:
| Plan | Price | Quota | Best for |
|---|---|---|---|
| Free | $0/mo | Free trial | Kicking the tires |
| Standard RECOMMENDED | Pay-as-you-go/mo | Unlimited usage | Solo devs · small teams |
| Enterprise | Custom | SLA · dedicated support | Teams & agencies |
Supported models
2 models across major vendors.
Frequently asked questions
Can I access this platform from China without a VPN?
Most relay stations are accessible from Chinese ISPs. Check our review for specific routing details.
What payment methods are accepted?
Payment options vary by platform. Some accept Alipay/WeChat Pay, others are USD/crypto only.
How does this compare to using OpenAI directly?
Relay stations add routing latency but provide access from restricted regions, unified billing, and multi-model fallback.
Is my API key safe?
Keys are encrypted at rest. Most platforms support per-project scoping and IP allow-lists.
Should you use DoroAI?
Codex budget users