#FAQ

For failing requests start with Troubleshooting; for unfamiliar terms read Core Concepts.

#I lost my key — can I recover it?

No. Keys are shown in full only once at creation; the server stores only a hash. Delete the old key on the API Keys page and create a new one.

#Can one key use several groups?

Yes. Enable the composite-key switch at creation, assign each group a prefix, and call models as prefix/modelID. See Composite Keys.

#Why does the same model appear twice?

It is split into two groups by protocol format (OpenAI / Anthropic). Pick the one matching your client — the wrong one returns 403. See API Endpoints.

#What is the deduction order? Does unused subscription quota survive expiry?

Subscription first, then balance: the subscription expiring soonest is consumed first, and the balance takes over once subscriptions run out. Unused subscription quota is cleared at the end of its cycle; balance never expires. See Billing.

#Are model plaza prices base or final?

Final. Prices in the model plaza already include the group multiplier — what you see is what you pay.

#Which models are supported?

See the live list in the model plaza. Remember each key can only use models within its bound group.

#Is 1M ultra-long context supported?

Yes. Some models support million-token-scale context, priced by context tier: one price in the base tier, a higher tier beyond. Which models support it follows the model plaza labels; tier rules are in Billing.

#The network is unstable — are there alternate endpoints?

Yes. Besides the default endpoint there are CN2-GIA routes in the US West and Germany; all three are fully equivalent. See API Endpoints.

#What is concurrency and how do I raise it?

The account-level cap on parallel requests, shown in your profile. It can be raised with redemption codes.

#Responses are slow — can I speed them up?

Enable Fast Mode on the key to force the high-priority lane at roughly 2× cost. Slow first tokens on long-reasoning models are a model property — paying more does not skip the thinking phase.

#Who pays for team members' usage?

The owner. All team-key spending is deducted from the owner's subscriptions and balance; members' own balances are untouched. Owners can set daily/weekly/monthly limits per member on the team page.

#Which client for image generation?

On desktop, Cherry Studio — pick the image model in a conversation and send a prompt, see the Cherry Studio guide; on Android use RikkaHub.

#How do I get an invoice?

Invoices are not self-service; contact support to have one issued manually, see Invoices.

#Does ByteSeek have other products?

There is an independent web chat platform with separate accounts and billing. See the documentation home for a comparison.

#Enterprise contact?

Business inquiries: SourceOne@mail.byteseek.ai — describe your scale and scenario and someone will follow up.