IKunCode routes only 2 models: GPT-4o and Claude 3.5 Sonnet. That’s it — no Gemini, no DeepSeek, no Mistral. For a relay station operating in 2026, this is a skeleton lineup.
Model-by-Model Breakdown
GPT-4o
- Context window: 100,000 tokens (matches OpenAI’s standard)
- Stability: Acceptable on most channels, but the 94% uptime means you’ll hit ~6 hours of downtime per month. Not terrible, but not great.
- Speed: No hard numbers in the data, but reverse-proxy setups from this tier typically add 200-500ms latency over direct API calls.
- Worth using? Only if you need Chinese payment methods (Alipay/WeChat Pay) and can’t get GPT-4o elsewhere. The pricing is “slightly above average” per the cons — expect to pay more than OpenRouter or direct OpenAI access.
Claude 3.5 Sonnet
- Context window: Same 100k token limit
- Stability: Shares the same infrastructure, so same 94% uptime. Some channels may have intermittent issues — the data flags this explicitly.
- Speed: Reverse-proxy Opus 4.6 is listed at ¥2(input)10(output) — that’s ¥2 input / ¥10 output per million tokens. Sonnet pricing isn’t separately specified, but expect similar markup.
- Worth using? The reverse-proxy Opus 4.6 mention is odd — it’s not in the Models list, so I can’t verify it. Stick to Sonnet if you need it, but don’t expect reliability.
Pricing
| Model | Context Window | Uptime | Notes |
|---|---|---|---|
| GPT-4o | 100k tokens | 94% | Slightly above average pricing |
| Claude 3.5 Sonnet | 100k tokens | 94% | Intermittent channel issues |
| No promo codes available. No refund policy stated. Min recharge not specified — you’ll likely need to ask in the QQ group. |
Pros & Cons
Pros
- Dedicated status monitoring page — you can see live uptime without guessing
- Active QQ community for support (though that’s also the only support channel)
- Reverse-proxy Opus 4.6 pricing is transparent: ¥2 input / ¥10 output per million tokens
- Accepts Alipay and WeChat Pay — useful if you don’t have international credit cards Cons
- Only 2 models — this is a stripped-down relay
- 94% uptime is below average for relay stations in 2026 (most target 99%+)
- Support is QQ group only — no ticket system, no email, no Discord
- “Slightly above average pricing” means you’re paying more for less model variety
- Some channels have intermittent issues — no SLA or compensation mentioned
Verdict
IKunCode is a niche relay for a specific use case: Chinese developers who need GPT-4o or Claude 3.5 Sonnet, can pay with Alipay/WeChat Pay, and don’t mind QQ-based support. The 94% uptime and 2-model lineup make it a poor choice for production workloads. The status page is a nice touch, but it doesn’t compensate for the limited selection and above-average pricing. If you need more models or better reliability, look elsewhere. If you’re stuck in China without a VPN and just want a quick way to hit GPT-4o or Sonnet, IKunCode works — barely.
FAQ
Q: Does IKunCode support streaming responses? A: Not explicitly stated in the platform data. Most reverse-proxy relays support streaming by default, but with only 2 models and 94% uptime, expect occasional interruptions mid-stream. Q: Can I get a refund if the service goes down? A: No refund policy is specified. Given the QQ-only support model, assume all sales are final. Q: How do I get help if the QQ group is down? A: You don’t. There’s no alternative support channel. The status page will show if the relay itself is up, but if the QQ group is unavailable, you’re stuck waiting.
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.
