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

# Dedicated Service Mode

> Models running on Hostlinku’s own infrastructure-DeepSeek, GLM, and Kimi-with a context size of 1 million tokens.

**Dedicated** is Hostlinku’s third service mode. Unlike **Performance** and **Low-Cost**, which forward requests to upstream partners, the Dedicated model runs on **Hostlinku’s own infrastructure**.

One API key = one Service Mode

This mode is enabled when your API key is set to "Dedicated" by an admin. To use another mode at the same time, create a separate API key-Service Mode is tied to the key, not to the request.

## Available Models

Nine models, three families:

| Family | Model | Context |
| - | - | - |
| DeepSeek | `deepseek-v4-1-flash`, `deepseek-v4-1-flash-turbo`, `deepseek-v4-1-flash-hyper` | 1,000,000 tokens |
| GLM | `glm-5-3`, `glm-5-3-flash`, `glm-5-3-flash-hyper` | 1,000,000 tokens |
| Kimi | `kimi-k3`, `kimi-k3-turbo`, `kimi-k3-hyper` | 1,000,000 tokens |

The `-turbo` and `-hyper` variants are faster/larger configurations of the base model. Models not listed here are **not** available in Dedicated mode-use Performance or Low-Cost instead.

## Endpoints and Authentication

The base URL does not change between modes. All three API dialects are supported:

| Dialect | Endpoint | Authentication header |
| - | - | - |
| OpenAI Chat | `POST {base}/v1/chat/completions` | `Authorization: Bearer <API_KEY>` |
| OpenAI Responses | `POST {base}/v1/responses` | `Authorization: Bearer <API_KEY>` |
| Anthropic Messages | `POST {base}/v1/messages` | `x-api-key: <API_KEY>` |

Simply point the app to the Hostlinku base URL and use the Dedicated mode API key-there are no additional parameters for selecting the mode.

## Price

Selling price = base price of the model × Dedicated mode multiplier, billed per **1M tokens** (1 kredit = \$1). The figures per model are listed in the [Model Catalog](https://docs.hostlinku.com/en/models.md) - select the **Dedicated** tab to view the prices applicable to your key. Example: `glm-5-3` currently costs kredit1.21 inputs and
kredit3.80 output per 1M token.

## Prompt Caching

The Dedicated model uses **the provider's auto-cache**. You don’t need to send `cache_control`: send the same prefix within the cache’s lifetime, and the tokens read from the cache will appear as `cached_tokens` (`/v1/chat/completions`) or `cache_read_input_tokens` (`/v1/messages`). See [Save Money with Prompt Caching](https://docs.hostlinku.com/en/prompt-caching.md) for instructions on how to view the metrics in the console.

## Limitations

* There are only nine models in the table above; there is no automatic fallback to another mode.
* The `-chat` variant (for coding/agentic use only) is not available for sale in the public catalog.
* Tool calling, streaming, and `stream_options.include_usage` are supported. Vendor-specific features
  (e.g., web search) depends on the model and is not guaranteed to be available.


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