Advanced system settings

What is EPay

EPay is a third-party checkout that aggregates channels such as Alipay and WeChat Pay. When a user tops up, the form fields are obtained via `POST /api/user/pay`, the browser POSTs them to the EPay gateway to complete payment, and an asynchronous callback updates the quota.

EPay configuration

Fill in the payment address, merchant ID/key and callback address. After saving, check on the Wallet page whether `GET /api/user/recharge/info` returns pay_methods.

Stripe configuration

Configure the Stripe Secret Key, Webhook Secret and related values in the Payment settings tab. User top-ups go through `POST /api/user/stripe/pay` and redirect to Checkout.

Other payment methods

Channels such as Creem are supported (depending on option toggles). The minimum top-up and callback path for each channel are whatever the API returns.

Top-up methods

Use JSON to customize the top-up method buttons shown in the frontend (name, color, type); type maps to the EPay type parameter.

Field reference

See the “EPay parameters” and “Server address” tables below. After changing the key, run a small test top-up again.

Top-up amounts

Configure fixed tiers (such as 10/50/100 USD) or allow custom amounts. Together with MinTopUp and Price, these constrain what users can enter.

Custom top-up amount option

When enabled, the Wallet page shows an input for any amount; when disabled, only the preset tiers are shown.

Top-up bonuses

Configure a top-up bonus ratio per group (TopUpGroupRatio). For example, a user in the vip group who tops up 100 USD receives quota worth 110 USD.

EPay parameters

Global rate limit

Set site-wide QPM/QPD caps in the “Rate limit settings” tab; requests over the limit return 429. Use this to protect the overall capacity of the gateway.

Rate limits by user group

Set independent thresholds for each group; these take precedence over the global limit. Groups without their own setting use the global values.

Group rate limit example

JSON example: `{"default":{"qpm":60},"vip":{"qpm":300}}`. Model-level limits can be refined by model name in the same tab.

Ratio system overview

EasyAPI calculates quota consumption from a combination of model_ratio, completion_ratio, group_ratio, cache_ratio and more; model metadata and the option table both take effect.

How quota relates to ratios

User quota is a unified currency unit. Different models and groups use ratios to convert it into equivalent token or per-call consumption.

Quota formula

Usage-based: cost ≈ (prompt_tokens × model_ratio + completion_tokens × model_ratio × completion_ratio) × group_ratio × base unit price. Per call: cost = model_price × group_ratio.

Usage-based billing (tokens)

quota_type = 0; input and completion tokens are weighted by separate ratios. See the pricing docs for details.

Per-call billing (fixed price)

quota_type = 1; each call deducts model_price regardless of the token count.

Audio models (special handling)

Some audio models are billed by duration or characters; the meaning of model_ratio follows the configuration in Model management.

Pre-deduction and post-deduction

Pre-deduction reserves an estimated quota before the request and settles the difference based on actual tokens afterwards; post-deduction charges once after the response completes.

Billing types

Model ratio settings

Controls the base input ratio (model_ratio) of each model. Displayed price (USD) ≈ model_ratio × 2 / 1M tokens. See Model management for details.

Completion ratio settings

Group ratio settings

The group ratio (group_ratio) sets the billing discount for the same model under different API key groups, and matches the ratio field in Group management.

{
  "default": 1,
  "vip": 0.8,
  "svip": 0.6,
  "test": 0
}

Visual ratio editor

The “Ratio settings” tab offers table-based editing with search, sorting and bulk save, so there is no need to write JSON by hand. Web console support is still being integrated.

Models without a ratio

When channel sync or manual creation adds models that have no model_ratio yet, the behavior is controlled by the option accept_unset_model_ratio_model:

• Off (default): calls are rejected and the log shows “model ratio not set”, preventing incorrect billing.
• On: calls are allowed and billed at the default ratio of 1 or the upstream value; useful for onboarding new models quickly and repricing them later.

Keep this off in production and enable models only after configuring them in “Model management” or through ratio sync.

Upstream ratio sync

Pull the latest model_ratio from connected upstream channels to reduce manual maintenance.

1. Call GET /api/ratio_sync/channels to list channels that support sync
2. Select a channel, then call POST /api/ratio_sync/fetch to pull its ratios
3. Review the differences in the visual table or JSON, then save
4. If needed, use POST /api/setting/rest_model_ratio to restore the default ratios

FAQ

Quota calculation examples

The examples below show how weights are calculated. The final quota also depends on the site's base unit price and display currency; the exact charge is shown in Usage records.

Chat app configuration

Controls the default model and visibility of the Playground and chat page; saving affects `/console/playground` and the demo chat entry.

Chat integration variables

The following placeholders can be used in one-click import URL templates. The frontend replaces them with the current user's API key and the site's API address when opening the third-party app:

• {apiKey} / {token}: the currently selected EasyAPI API key (starts with sk-)
• {baseUrl} / {openAIUrl}: the site's API base URL (NUXT_PUBLIC_API_BASE_URL)
• {openAIBaseUrl}: {baseUrl}/v1, for OpenAI SDK-compatible clients
• {siteName}: the site name, used as the title in some clients

Chat app integration

The “Chat app integration” section on the API keys page supports one-click import into clients such as ChatGPT Next Web, Lobe Chat and OpenCat. See “Chat app integration” in the user guide for how each app's parameters are assembled.

• Create and select the target API key on the API keys page before importing
• Import links open in a new tab, so the full key is not leaked into the current page's URL history (some clients may still briefly show it in the hash)
• For manual integration, use Authorization: Bearer {apiKey} and {openAIBaseUrl}

For the full steps, see the Chat app integration user guide.

Midjourney configuration

Controls MJ proxy behavior, notifications and account filtering. Configure a Midjourney channel in “Channel management” first.

Drawing billing

Image models such as Midjourney are usually billed per call (quota_type=1). Set model_price in Model management; the group ratio (group_ratio) also applies.

• Actions such as imagine / upscale / variation are each charged once or according to the model configuration
• Quota is pre-deducted when a task is submitted; failed tasks are refunded according to the gateway policy
• Users can view mj_id, progress and charges at `/console/tasks`
• See “Task center” and the per-call billing notes in “Pricing” in the user guide

Related docs: Task center, Pricing.

Dashboard - basic settings

Controls the dashboard's default time range and whether key metric cards such as quota and request volume are shown.

Dashboard - chart settings

Configure data granularity and displayed series for line and bar charts, helping operators analyze peak hours and model distribution.

Dashboard - advanced options

Extended capabilities for Root/admins, such as raw log export and cross-user aggregation.

• Data export: export CSV by time range (available fields depend on the console)
• Admin view: see site-wide totals; regular users only see their own data
• Cache refresh: after changing dashboard options, clear the CDN/browser cache so changes take effect immediately

Model display settings

Affects what is shown on the Models page `/models` and in console dropdowns.

Model behavior settings

Globally affects API gateway routing, retries and how models without a ratio are handled.

Model sync settings

Automatically discover model names from channels and write them to the model table; can be combined with ratio sync.

• After enabling channels in “Channel management”, trigger “Sync model list from channels” in model settings
• Sync only updates model names and metadata; ratios must be configured separately or pulled via `POST /api/ratio_sync/fetch`
• Model mapping (model_mapping) is configured per channel and is not duplicated in this tab

For maintaining channels and model lists, see Channel management and Model management.

Basic operation settings

Controls new user registration, initial quota and the site's operating mode.

Top-up settings

Works with the Payment tab to control top-up capabilities on the Wallet page `/console/wallet`.

Redeem code settings

Controls user self-service redemption and bulk code generation by admins.

• For bulk generation by admins, see the “Redeem code management” docs (admin API: `/api/redeem_codes`)
• User redemption entry: `/console/wallet`, API `POST /api/user/redeem`

For admin operations, see Redeem code management.

Home page settings

Content for the marketing home page and landing pages, returned read-only by public APIs.

Other feature settings

Miscellaneous items such as OAuth, notifications and the footer. For OAuth providers, see the “Custom OAuth” docs (if enabled).