> ## Documentation Index
> Fetch the complete documentation index at: https://docs.kiosapi.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Choosing a Token Group

> Pick the right group when creating an API key so your requests route to the models you actually want to use.

# Choosing a Token Group

Every KiosAPI API key (token) belongs to a **group**. The group decides which models the key can access and at what price multiplier. Picking the right group **before** creating the key is the most important step — a key created in the wrong group simply cannot call the models you want.

## How groups work

* Each group unlocks a specific set of models (for example `Gemini`, `Claude-Kiro`, `Codex-Pro`, `Free`).
* Each group has its own price multiplier on top of the base model price — some groups are heavily discounted.
* A key can only call models enabled for its group. If you request a model outside your group, you get an error instead of a response.

## Step-by-step: create a key with the right group

1. Open the [KiosAPI Dashboard](https://kiosapi.com) and go to **Console → Tokens** (Token Management).
2. Click **Add Token**.
3. Give the token a name, quota, and expiry.
4. **Select the group** that matches the model family you plan to use — this is the critical step.
5. Save and copy the key (starts with `sk-`). The group is bound to the key from this moment.

> [!WARNING]
>
> You cannot call a model with a key whose group does not include it. Before creating the key, check which group serves your target model on the [Models & Pricing](https://kiosapi.com/pricing) page — the pricing table shows the group for every model, along with its live price.

## Groups currently on KiosAPI

The live, authoritative list is always the [Models & Pricing](https://kiosapi.com/pricing) page. As a quick orientation, the groups below show which model families they typically unlock and their price multiplier:

| Group | Typical models | Multiplier | Notes |
| - | - | - | - |
| `default` | General pool (GLM free tier, Nemotron, etc.) | 1.0 | General-purpose fallback |
| `Free` | Free models (Agnes, Atria, GLM flash-free, gpt-oss-20b, ...) | 0 | Free — requires Telegram verification, see [Free Models Guide](/getting-started/free-models.md) |
| `Claude-Kiro` | Claude Opus / Sonnet series | 0.05 | For Claude Code and Anthropic workloads |
| `Codex-Pro` | gpt-5.5, gpt-5.6-sol, gpt-6-astra | 0.04 | For Codex CLI coding workflows |
| `Codex-Plus` | gpt-5.6-sol, gpt-5.6-terra, gpt-6-astra | 0.02 | Cheaper Codex tier |
| `Gemini` | Gemini flash series | 0.05 | For Gemini CLI |
| `Grok` | grok-4.5 / 4.6 / 4.7 | 0.03 | |
| `DeepSeek` | deepseek-v4 series | 0.2 | |
| `DeepSeek-Promo` | deepseek-v4-flash | 0.01 | Promo tier |
| `GLM` | glm-5.2 / 5.3 series | 0.2 | |
| `China-Models` | Kimi, Qwen, MiniMax, HY | 0.2 | |
| `Image` | gpt-image-2 / 2.5, grok-imagine | 1.0 | Image generation models |

> [!NOTE]
>
> Multipliers and model lists change over time. Always treat the [pricing page](https://kiosapi.com/pricing) as the source of truth — the table above is an orientation snapshot, not a price guarantee.

## Common recipes

| I want to use... | Create the key in group... |
| - | - |
| Claude Code | `Claude-Kiro` |
| Codex CLI | `Codex-Pro` or `Codex-Plus` |
| Gemini CLI | `Gemini` |
| Free models at no cost | `Free` (after Telegram verification) |
| Image generation (gpt-image, grok-imagine) | `Image` |
| A bit of everything | `default` |

## Changing groups

A token's group is fixed at creation. To use another group, create a new token with that group — existing tokens keep their original assignment. Keeping one token per group (one for coding, one for free models, one for images) is the recommended pattern: it keeps quotas and revocation independent.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.