The 30-second summary
+ What we liked
- Widest model support: Claude, GPT, Gemini, DeepSeek, Kimi, Grok, GLM, Qwen
- Pay-per-call and pay-per-token both available
- Daily check-in rewards
- ¥1 minimum recharge
− What we didn't
- Cheapest channel consistently dilutes (mixes lower-quality models)
- Some channels show variable quality day-to-day
- Support response time varies
In-depth review
LinkAPI Review: A Broad Model Hub with Pay-As-You-Go Flexibility
If you’re a developer in China needing access to a wide range of LLMs without a VPN, LinkAPI presents an interesting proposition. It’s not the most stable or highest-quality relay station, but its sheer breadth of model support and flexible pricing make it a strong candidate for experimentation and multi-model workflows.
Models and API Compatibility
LinkAPI’s biggest strength is its model selection. It supports Claude, GPT, Gemini, DeepSeek, Kimi, Grok, GLM, and Qwen – one of the widest lineups I’ve seen. This is a major advantage if you’re building apps that need to switch between different providers or want to compare outputs across models.
The API is OpenAI-compatible, which means you can plug it into most existing tools (LangChain, Open WebUI, etc.) with minimal fuss. Just swap the base URL and API key.
Max token context: 100,000 tokens. This is decent for most use cases, though it won’t handle very long documents or full codebases.
Pricing and Payment
LinkAPI offers two pricing models: pay-per-call and pay-per-token. This flexibility is rare and useful. For short, frequent queries, pay-per-call can be cheaper. For long, streaming outputs, pay-per-token is more predictable.
| Feature | Detail |
|---|---|
| Minimum recharge | ¥1 |
| Pricing models | Pay-per-call and pay-per-token |
| Free tier | No |
| Daily check-in rewards | Yes (free credits) |
The ¥1 minimum recharge is the lowest I’ve seen. This makes it trivial to test the service without commitment. The daily check-in rewards are a nice bonus for regular users, though they won’t cover heavy usage.
Uptime and Quality
Uptime: 96% – This is below the industry standard (most relay stations aim for 99%+). In practice, you’ll notice occasional outages, especially during peak hours in China. The safety rating is 3/5, which reflects the variable quality of the underlying channels.
The biggest caveat: the “cheapest channel” often dilutes quality by mixing lower-tier models. You might request GPT-4o and get a less capable variant. Quality also fluctuates day-to-day on certain channels. This is a dealbreaker if you need consistent, high-quality outputs for production.
Developer Experience
Setup is straightforward: create an account, recharge ¥1, generate an API key, and start hitting endpoints. The dashboard is functional but not polished. Support response time varies – sometimes quick, sometimes slow.
For local testing, it works fine. For production, the variable quality and 96% uptime are concerning. I’d recommend using it as a fallback or for non-critical workloads.
Pros & Cons
Pros
- Widest model support: Claude, GPT, Gemini, DeepSeek, Kimi, Grok, GLM, Qwen
- Pay-per-call and pay-per-token both available
- Daily check-in rewards
- ¥1 minimum recharge
Cons
- Cheapest channel consistently dilutes (mixes lower-quality models)
- Some channels show variable quality day-to-day
- Support response time varies
Verdict
LinkAPI is a good choice for developers who need broad model access and want to pay as they go, but it’s not for production use. The 96% uptime and variable quality mean you can’t rely on it for mission-critical applications. However, for prototyping, testing, or multi-model experiments, the ¥1 minimum recharge and wide model support make it worth a try.
If you need consistent, high-quality outputs, look elsewhere. If you want to explore many models without breaking the bank, LinkAPI is a solid option.
FAQ
Q: Can I use LinkAPI without a VPN in China?
A: Yes, LinkAPI is designed for Chinese developers and works without a VPN. It routes API calls through local servers.
Q: How does the “cheapest channel” dilution work?
A: When you select the cheapest pricing tier, LinkAPI may route your request through lower-quality model variants. For example, you might request GPT-4o but receive a less capable version. Stick to higher-priced channels for consistent quality.
Q: What happens if I run out of credits mid-request?
A: The API will return an error. Since the minimum recharge is only ¥1, you can quickly top up. There’s no auto-recharge feature, so monitor your balance if you’re running heavy workloads.
Pricing breakdown
LinkAPI offers competitive pricing for developers. Here's the breakdown:
| Plan | Price | Quota | Best for |
|---|---|---|---|
| Free | $0/mo | Limited | 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 LinkAPI?
Users who need access to many different models