FAQ

Frequently asked questions about Claude Code and QCode.cc

Need an API key? Sign up →·Endpoints & base URLs →

QCode.cc Platform¶

Q: Where do I get my API Key?¶

A: Log in to the QCode.cc Dashboard and create or view your API Key on the "API Key Management" page.

Q: What prefix does the API Key use?¶

A: QCode.cc API Keys start with cr_, e.g. cr_xxxxxxxxxxxxxx.

Q: Why is my API Key not working?¶

A: Common causes:

1. Copy error: Check for extra spaces or missing characters
2. Subscription expired: Check your account status on the Dashboard
3. Environment variable not applied: Reopen your terminal or run source ~/.zshrc
4. Wrong API endpoint: Make sure ANTHROPIC_BASE_URL is set to https://api.qcode.cc/api

Q: My API Key leaked — how do I replace it?¶

A: It depends which key:

• Main API Key (starts with cr_): reset it yourself in the console — on the key card, click «⋯» at the right of the «API Key:» line → «Reset API Key…». You get a new key, the old one stops working within 1 minute, and your quota, usage and expiry date stay the same. Each reset is charged a small fee (the amount is shown in the console), and each key can be reset at most once every 3 days. If you only forgot the key, no reset is needed — click the 👁 icon next to it. A few keys can only be reset with help from live chat or hi@qcode.cc. Also follow Security Best Practices to check whether the key reached your Git history or logs.
• Open API access token (starts with qot_, personal): read-only, generated in the console under Open API (/dashboard/openapi-token); reset or revoke it there. The old token stops working immediately. See Dashboard Guide.
• Partner access token (starts with rsk_, resellers only): issued in the Partner Dashboard under «Open API / Access Token», not in the personal console. Click Reset there. The old token stops working immediately. See Partner Open API.

Usage & Plans¶

Q: How do I check my current session usage?¶

A: Two ways:

Dashboard (recommended): Log in to the Dashboard and view model call counts, cost breakdown, and plan consumption on the "Usage Statistics" page.

Quick check in CLI: Use the /cost command to see a summary of the current session:

/cost

Dashboard billing is based on the qcode.cc/models pricing table, which is more accurate than what the CLI shows.

Q: Is there a usage cap on the Max x20 account resources for Ultimate and Enterprise Team plans?¶

A: Yes. Each Max x20 account has a cap of $4,000 per month (monthly reset, unused quota does not roll over; actual consumption is typically $3,500–$3,900). We schedule through an account resource pool, so the vast majority of users will never hit this limit. For unusually high usage requirements, please contact support.

Q: What are the concurrency limits and time-window limits for each plan?¶

A:

Concurrency limits:

• Personal plans: 5 concurrent requests per API Key; Ultimate: 10 concurrent requests per API Key
• Pay-as-you-go Packs: 5 concurrent requests per API Key
• Enterprise Team: 35 concurrent requests per account resource, 3 users per account recommended, up to 7 sharing one account

4-hour time-window limit:

• Personal plans: fully removed as of July 2026 — no time-window limit of any kind; the cap that applies is your daily limit (reset at 00:00 Beijing time), or the monthly quota on Ultimate
• Enterprise Team: no time-window limit, supports continuous development

Q: Not sure which tier to buy — can I upgrade if I start too small?¶

A: Yes. If you are unsure how much quota you need, start with the Starter plan ($9/mo). Whenever you are ready, buy a higher-quota plan and email hi@qcode.cc to close the old one — its unused time is credited back to your QCode balance pro rata, handled within 1 business day. Teams and resellers can also validate the integration on Starter first and upgrade afterwards.

Q: What is the difference between a Pay-as-you-go Pack and a personal plan?¶

A: A personal plan has a daily quota that resets at 00:00 Beijing time and suits daily use; a pack is a total quota for its validity period (Small / Medium / Large / XL Pack, all valid for 7 days), used until it runs out, and suits stretches of use. The key works the same way, with concurrency 5 and the same models. You cannot buy a pack while you have an active personal plan (when the daily quota of the personal plan is not enough, use a booster pack). For the full comparison see How to Choose a Plan, and for the rules see Pay-as-you-go Packs.

Q: What happens when the quota runs out?¶

A: The response is HTTP 402 with error.type = insufficient_quota, and code tells you which quota: daily_cost_limit_exceeded is the daily quota, which resets at 00:00 Beijing time (a booster pack also helps); total_cost_limit_exceeded is a total quota (Pay-as-you-go Packs, the Ultimate monthly quota), and for a pack you simply renew; the weekly limit for Claude models also returns 402. An expired key returns 401, not 402, and renewing a pack restores it. See Error Codes.

Q: Are refunds available?¶

A: We sell digital network goods. Once activated, account resources are occupied, so refunds are not supported. Pay-as-you-go Packs are not refundable either, and remaining quota is void after expiry.

The one exception is upgrading: after buying a higher-quota plan, email hi@qcode.cc and the unused time on the old plan is credited back to your QCode balance pro rata (to your balance, not to the original payment method).

Balance transfer alternative: Transfers between QCode members are supported. Once you find a buyer, send an email to hi@qcode.cc with the transfer amount and the recipient's QCode account email address. Transfers are completed within 1 business day.

Refunds may be granted on a case-by-case basis only in the following situations: a platform outage causing more than 24 hours of service interruption; the platform unilaterally terminates service while a balance remains; or a mistaken top-up is reported within 24 hours with no usage having occurred.

Q: I have a Claude Pro / ChatGPT Plus subscription — can I use it with QCode?¶

A: No. A subscription is membership in the official apps and includes no API quota; QCode bills per token. The two are independent systems and can coexist — details in Subscriptions vs API Keys vs QCode Keys.

Q: Which tools are supported?¶

A: Claude Code, Codex, IDE / CLI extensions and desktop clients all have setup pages under "Connect Your Tool"; for which protocol each tool speaks and whether Claude models work with it, see the Tool Compatibility Overview.

Q: Do you support corporate bank transfer?¶

A: The Enterprise Team plan supports corporate bank transfer and a formal service agreement. See the Enterprise Guide.

Q: I got daily_cost_limit_exceeded (HTTP 402) — what now?¶

A: The daily limit is the cap on API spend per day, and it resets at 00:00 Beijing time. Two ways out:

1. Buy a booster pack: purchase it yourself in the console to extend today's quota immediately. Booster packs reset separately at 02:00 Beijing time the next day — two hours after the daily limit, which leaves a buffer if you work past midnight.
2. Raise the daily limit: adjust your plan at qcode.cc/pricing.

This arrives as HTTP 402 with error.type = insufficient_quota and code = daily_cost_limit_exceeded (see Error Codes).

The 4-hour rolling window limit was removed from all personal plans in July 2026, so this message can only come from the daily limit, never from a window limit.

Q: What does "total cost limit" (total_cost_limit_exceeded) mean?¶

A: Your total quota is used up: HTTP 402 with error.type = insufficient_quota and code = total_cost_limit_exceeded. The total quota applies to keys billed against a total quota, for example Pay-as-you-go Packs and the monthly quota of the Ultimate plan. The amount in the message's parentheses (for example $100) is the key's cumulative limit, not what is left.

How it recovers depends on your plan's rules: for a pack, renewing is enough; for example the Ultimate monthly quota is topped up at each monthly reset. If it ran out before the reset, or you are not sure which kind applies to you, email hi@qcode.cc or use live chat. This kind of 402 does not go away by resending, so there is no point in a retry loop. See Error Codes.

Q: Are cloud servers and the AI service the same thing? Do I have to buy an AI plan as well?¶

A: They are separate products, sold independently and sharing only one account balance. A cloud server is an overseas machine with a dedicated public IPv4 and full root access, charged for the term you pick (1, 3, 6 or 12 months); the AI service is billed per token. You can buy a cloud server without any AI plan: install Claude Code or Codex on it and call the models with your QCode key. Picking the machine is covered in Cloud Servers: Picking, Buying and Using, the server-side billing and refund rules in Cloud Server Billing and Rules, and the setup plus long-running agents in Run agents on a cloud server.

Quick Reference¶

Q: I just bought a plan — how do I start using it?¶

A: Three steps, about 60 seconds:

1. Get the key: sign in to the console → "API Key Management" and copy the key that starts with cr_.
2. Configure: run the one-click setup script, or set ANTHROPIC_BASE_URL and ANTHROPIC_AUTH_TOKEN by hand (see Endpoints & API Formats).
3. Pick your tool: find your IDE or CLI under Connect Your Tool and follow that page.

One key drives Claude Code, Codex, Gemini and the Chinese models at the same time, all sharing the same plan quota.

Q: How do I install Claude Code?¶

A: Install globally via npm, then configure the environment variables:

npm install -g @anthropic-ai/claude-code

Detailed steps (including instructions for all three platforms) → Installation Guide

Q: How do I configure the QCode.cc proxy?¶

A: Set the following environment variables:

export ANTHROPIC_BASE_URL="https://api.qcode.cc/api"
export ANTHROPIC_AUTH_TOKEN="cr_your_api_key"

We recommend adding these to your shell config file (~/.zshrc or ~/.bashrc). See → Installation Guide

Q: How do I enter Plan Mode?¶

A: Press Shift+Tab until the mode indicator shows plan (three times from the default auto mode, twice from manual default mode). You can also start with --permission-mode plan, or set the default mode in .claude/settings.json:

{ "permissions": { "defaultMode": "plan" } }

In Plan Mode, Claude analyzes your request without executing any actions.

Q: How do I switch models?¶

A: Use the /model command:

/model opus
/model sonnet

Q: How do I have Claude commit or review code for me?¶

A: Use /commit to generate a commit and /review to perform a code review. You can also just tell Claude directly: "commit these changes" or "review this PR".

Q: How do I report a bug?¶

A: Visit GitHub Issues and provide reproduction steps along with any relevant error information.

Q: Can Claude generate images?¶

A: Claude models can read images but not generate them (for image input see Image Input).

To generate images use gpt-image-2 — it shares your existing cr_ key, is fully OpenAI-compatible, and supports both text-to-image and image editing. See gpt-image-2 Image Generation and Editing.

More Help¶

Can't find what you're looking for? Check these topic-specific docs:

Contact support: Use the live chat in the bottom-right corner of the website, or email hi@qcode.cc

Enterprise →