Common Errors & Fixes

Don't panic when you see an error — below they're grouped by "the message you're seeing," with the cause and what you should do for each one. The fastest way is to search for a keyword with Ctrl/⌘ + F in your browser.

• 401 / invalid token
• Insufficient balance / insufficient quota / INSUFFICIENT_BALANCE
• No available channel / no available channel under the current group
• Model not found / model_not_found
• Stream disconnected / Connection error / high load
• Request timeout / timeout
• reasoning / reasoning effort not taking effect
• Console page 500 / blank screen / everything errors when clicked
• Deep-dive guides by error
• Still stuck?

401 / invalid token

Cause: a problem with your API Key (token). Most often it's one of these:

• The Key was copied wrong (an extra space, a missing character)
• The Key has been deleted or disabled in the console
• It's in the wrong place in your client, or it's missing the Bearer prefix

What to do:

1. Go back to Console → Tokens and confirm this token still exists and its status is normal.
2. Copy it again and replace the whole string, making sure there are no spaces at the start or end.
3. Confirm base_url is also filled in correctly (see the guide for your client).

Insufficient balance / insufficient quota / INSUFFICIENT_BALANCE

Cause: your account balance, or this token's quota, has run out.

What to do: top up in the Console; or check whether this token has a quota cap set, and raise it.

Top-up not credited?

USDT / crypto top-ups must be paid in the exact amount shown on the page (paying too much or too little will not be credited automatically). If you paid the wrong amount, don't worry — take a screenshot of the payment record and contact support to have it handled manually.

No available channel / no available channel under the current group

Cause: the model you requested has no matching channel in the group your account belongs to. This commonly happens when you enter a model name we haven't listed, or one that isn't part of your plan.

What to do:

1. Open the Console and look at the list of currently available models, then pick one that's on it.
2. Make sure the model name is spelled exactly the same (case and hyphens all need to match).
3. If it's correct and you still get the error → it may be a permissions issue with your group; contact support to confirm.

Model not found / model_not_found

Cause: the model name is misspelled, or the model isn't in your group.

What to do: same as the previous item — go to the console, check it against the model list, and enter the name into your client exactly, character for character.

Stream disconnected / Connection error / high load

Cause:

• The upstream is busy during peak hours, or there's a momentary network hiccup
• Long-haul international links can be occasionally unstable depending on your region and route

What to do:

1. Just retry once — it usually works.
2. If it happens frequently → try a different model (different models use different upstreams).
3. If it keeps failing → take a screenshot and contact support, and we'll check the channel status on the backend.

Request timeout / timeout

Cause: usually a network problem on the route between your region and the gateway.

What to do: retry; if it's consistently slow, test from another network or check your route to byesu.com.

reasoning / reasoning effort not taking effect

Cause: the two APIs use different syntax, and mixing them up means it gets ignored.

API	Correct syntax
/v1/chat/completions	String: "reasoning_effort": "high"
/v1/responses (Codex, etc.)	Object: "reasoning": { "effort": "high" }

What to do: choose the right syntax for the API you're using. In Codex, just use the config option model_reasoning_effort directly — there's no need to write it by hand.

Console page 500 / blank screen / everything errors when clicked

Cause: having Google Translate turned on in your browser (auto-translating the whole page) conflicts with the console and causes errors when you click around the page.

What to do:

1. Turn off Google Translate for that page (the translate icon in the address bar → show original), then refresh.
2. We've already added protection on the server side, but we still recommend not using whole-page translation when browsing consoles like this.

Known and hardened

We've already added a compatibility fix for this issue. If you still run into it after updating, turning off translation and refreshing once should basically sort it out.

Deep-dive guides by error

Longer, code-heavy guides for the errors developers hit most:

• Anthropic API error 529 (overloaded) — meaning, retries and mitigation
• Claude API 429 vs 529 — rate limits vs upstream overload
• 401 authentication errors — key format, headers and base URL
• "No available channel" — token groups explained
• 404 model not found — naming and availability
• Streaming connection dropped — SSE debugging
• Request timeouts — SDK defaults and long-inference settings
• Quota, balance and 402 billing errors

Still stuck?

Send support a screenshot of the original error message plus the client and model you're using — it's much faster than describing it in words.