About Jizhi API

Jizhi API is a Claude / GPT / Gemini large language model API proxy platform. We integrate official interfaces from mainstream models into a unified Base URL with a single sk- key, saving you the hassle of overseas credit cards, account registration, and cross-border network issues. We offer pay-as-you-go pricing—pay only for what you use.

How to tell if a model is authentic

The market for proxy services is filled with mixed quality, and cases of models being swapped or "dumbed down" are not uncommon. Here are a few ways to verify for yourself:

1. Don't ask "Who are you?"—Identity can be spoofed; asking directly is meaningless.
2. Use capability testing: Use a few reasoning, math, or coding problems with definitive answers. The same model should maintain a stable performance level across multiple calls.
3. Check the response fields: Authentic responses will contain complete model, usage, stop_reason, and other fields; be wary if the structure is overly simplified.
4. Test long context: Feed in content close to the nominal length to see if the information in the middle is being quietly truncated.
5. Spot-check at different times: If it works normally during the day but gets "dumber" at night, it's likely shifting to cheaper channels during peak hours.
6. Be wary of prices that are too low: Unreasonably cheap prices usually mean costs were cut elsewhere.

What we do

Jizhi API insists on three principles:

• Authentic models—We provide exactly the model you request; no bait-and-switch.
• Verifiable—We provide native protocol endpoints and per-request logs, allowing you to reconcile models, routes, tokens, and costs.
• Stable and reliable—Multi-route redundancy with automatic switching if a single route fails.

Feel free to "test" us using the methods above at any time.

Contact Us

• QQ / WeChat: 915669723
• Email: 915669723@qq.com