Get started
Choosing a group
A group decides which line a key uses and which price it pays. The same model costs different amounts and comes from different sources in different groups. Every group, what its labels mean, and how to choose for development, production and failover.
A group is a setting on an API key: it decides which line the key's requests use and which price tier they are billed at. Most models are offered in several groups, each with its own price and source. Calls look exactly the same in every group, so switching groups means switching keys, not code.
All groups
| Group | Group ID | Labels | Description | Models |
|---|---|---|---|---|
| Standard | default | Lowest priceMixed sources | Covers every model at the lowest price; some GPT and Claude routes are non-official | 603 |
| Azure | azure | Cloud | Microsoft Azure OpenAI — official, well priced, enterprise-grade | 116 |
| Azure Plus | azure-plus | Cloud | Premium Azure OpenAI line with o-series reasoning models | 135 |
| OpenAI Direct | openai | Official | OpenAI's own API | 169 |
| OpenAI Direct Plus | openai-plus | Official | The fastest, steadiest OpenAI line, incl. o-series and realtime | 174 |
| Official · Enterprise | official | Official | Enterprise line at official list prices, sourced from OpenAI and Anthropic, stability first | 309 |
| Claude Direct | claude-official | Official | Anthropic's own API; new models may have limited concurrency at launch | 47 |
| Claude List Price | claude-list | Official | Anthropic official line billed at list price | 47 |
| Claude Plus | claude | Mixed sources | Cost-effective Claude line; includes non-official sources, prices may change | 40 |
| Claude Lite | claude-lite | Mixed sources | The cheapest Claude line, non-official sources | 27 |
| Gemini Plus | gemini | Official | Google Gemini official line with video understanding | 70 |
| Gemini High Tier | gemini-pro | Official | Higher-quota Google Gemini official line | 70 |
| China Models | cn | Lowest price | DeepSeek, Qwen, Kimi, GLM, MiniMax and more at special prices | 195 |
| China Models 2 | cn-2 | — | Backup line for Chinese models | 18 |
| Economy (per token) | economy | Mixed sources | Non-official sources, billed per token at a discount | 42 |
| Economy (per call) | economy-call | Mixed sources | Non-official sources, billed per call — good for long prompts | 54 |
| Stable Diffusion | sd | — | Stable Diffusion image models | 5 |
| Video Line 1 | video-1 | Mixed sources | Veo and Grok video generation; non-official sources | 6 |
18 groups, from the live price list.
What the labels next to a group name mean:
| Label | Meaning |
|---|---|
| Lowest price | The cheapest line among comparable groups |
| Official | Requests go to the model vendor's official API |
| Cloud | Requests go to a cloud provider's model service |
| Mixed sources | The line includes non-official sources; prices are lower, and prices and available models may change at any time |
One model, different prices per group
Live prices for the same model in each group — the gap can be several times:
| Group | Input ¥/1M | Output ¥/1M | Cache read ¥/1M |
|---|---|---|---|
| StandardLowest | ¥1.5 | ¥9 | ¥0.15 |
| Azure | ¥1.5 | ¥9 | ¥0.15 |
| Azure Plus | ¥2.25 | ¥13.5 | ¥0.225 |
| OpenAI Direct | ¥3.38 | ¥20.25 | ¥0.338 |
| OpenAI Direct Plus | ¥4.5 | ¥27 | ¥0.45 |
| Official · Enterprise | ¥8.21 | ¥49.27 | ¥0.821 |
gpt-5.4-mini · Prices in CNY, cheapest group for each model; updated Nov 12, 2026, 8:17 PM.
| Group | Input ¥/1M | Output ¥/1M | Cache read ¥/1M |
|---|---|---|---|
| Standard | ¥4 | ¥20 | ¥0.4 |
| Official · Enterprise | ¥21.9 | ¥109.5 | ¥2.19 |
| Claude Direct | ¥15 | ¥75 | ¥1.5 |
| Claude List Price | ¥24 | ¥120 | ¥2.4 |
| Claude Plus | ¥8 | ¥40 | ¥0.8 |
| Claude LiteLowest | ¥3.2 | ¥16 | ¥0.32 |
Claude Sonnet 5.5 · Prices in CNY, cheapest group for each model; updated Nov 12, 2026, 8:17 PM.
The model hub shows each model's cheapest group by default; a model's own page lists its price in every group.
The key decides the group
- You pick the group when you create or edit an API key in the console. A key belongs to exactly one group.
- A request cannot override the group. To use another group, use another key.
- Groups do not all offer the same models. If a key's group lacks a model, the request returns 503 with the code
model_not_found. Use a key from a group that offers the model, or check the model's page to see which groups have it. - Calling the models endpoint with a key returns exactly the models that key can use.
How to choose
| Situation | Suggestion |
|---|---|
| Development, testing, internal tools | A group labeled "Lowest price" — get the flow working first |
| Customer-facing production, or you care where requests go | A group labeled "Official" or "Cloud" |
| Critical workloads | Keep two keys in different groups and fail over to the second when the first keeps returning 5xx |
| Chinese models | The dedicated Chinese-model group, usually the cheapest |
Before going live, send real requests with a key in the group you plan to use, and check quality, speed and the charges under Usage before you settle on a production group.
One project can use several keys at once — for example chat in one group and image generation in another, each with its own spending limit. Usage in the console breaks the bill down by key.
Still stuck? Add our WeChat support JsnonoChat and send the request id from the error message, and we will look into it.