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

# Models Settings

> Configure platform gateway routing, toggle available chat models, and bring your own custom model endpoints.

## Overview

The **Models** tab allows you to manage network gateway behaviors, select which platform AI models are visible in your active chat picker, and connect custom OpenAI-compatible endpoints using your own API keys.

***

## 🌐 Platform Gateway

The **Platform Gateway** section handles automatic network routing checks to ensure optimal connection stability and performance across proxy or VPN shifts.

<Frame caption="Platform Gateway controls for setting automatic routing check intervals.">
  <img src="https://mintcdn.com/memverge-1b064d4d/9UyDf9zIDz_f9H7C/configuration/assets/desktop-models-gateway.png?fit=max&auto=format&n=9UyDf9zIDz_f9H7C&q=85&s=3c46637202579ea941067cef1be78029" alt="MemoryBox Platform Gateway settings view" width="860" height="381" data-path="configuration/assets/desktop-models-gateway.png" />
</Frame>

* **Current Route:** Displays your active regional endpoint routing (e.g., `US`).
* **Automatic gateway routing check:** Periodically refreshes platform routing while the application is running (helpful after enabling a VPN or modifying proxy settings).
* **Check interval:** Configures how often automated routing checks occur (Default: *Every 5 minutes*).
* **Check now:** Triggers an immediate manual network gateway refresh.

***

## 🤖 Platform Models

Toggle models on or off to customize which options appear in your chat model picker drawer. Toggling any row saves your changes immediately.

<Frame caption="The Platform models configuration interface listing available AI providers, capabilities, and pricing tiers.">
  <img src="https://mintcdn.com/memverge-1b064d4d/9UyDf9zIDz_f9H7C/configuration/assets/desktop-models-list.png?fit=max&auto=format&n=9UyDf9zIDz_f9H7C&q=85&s=8fc3b5c156c1a54ca0de2c468a8e0504" alt="MemoryBox Platform models management view" width="905" height="937" data-path="configuration/assets/desktop-models-list.png" />
</Frame>

### Available Model Providers

#### OpenAI

| Model             | Capability | Price     | Description                              |
| :---------------- | :--------- | :-------- | :--------------------------------------- |
| **GPT-5.6 Sol**   | Quality    | Very High | Flagship reasoning for hard problems     |
| **GPT-5.6 Terra** | Balanced   | High      | Balanced performance for everyday work   |
| **GPT-5.6 Luna**  | Fast       | High      | Fast assistant for everyday chat         |
| **GPT-5.4**       | Balanced   | High      | Strong reasoning for coding and analysis |
| **GPT-5.4 mini**  | Fast       | High      | Balanced performance for most tasks      |
| **GPT-4.1 nano**  | Fast       | Low       | Ultra-fast for quick replies             |

#### Anthropic

| Model                 | Capability | Price     | Description                                  |
| :-------------------- | :--------- | :-------- | :------------------------------------------- |
| **Claude Fable 5**    | Ultra      | Very High | Flagship intelligence for long-running tasks |
| **Claude Sonnet 5**   | Balanced   | High      | Strong all-rounder for coding and work       |
| **Claude Haiku 4.5**  | Fast       | High      | Near-instant replies for simple tasks        |
| **Claude Opus 4.7**   | Ultra      | Very High | Flagship reasoning for complex research      |
| **Claude Sonnet 4.6** | Balanced   | High      | Strong all-rounder for everyday work         |
| **Claude Opus 4.8**   | Ultra      | Very High | Flagship reasoning for hard problems         |

#### Google

| Model                     | Capability | Price  | Description                           |
| :------------------------ | :--------- | :----- | :------------------------------------ |
| **Gemini 3.1 Pro**        | Quality    | High   | Flagship multimodal with huge context |
| **Gemini 3.5 Flash**      | Balanced   | High   | Fast all-rounder for everyday work    |
| **Gemini 3.1 Flash-Lite** | Fast       | Medium | Lowest latency for quick queries      |
| **Gemini 3 Flash**        | Balanced   | Medium | Fast for questions and drafts         |
| **Gemma 4 26B A4B**       | Fast       | Low    | High-performance open model           |
| **Gemma 3 12B**           | Fast       | Low    | Lightweight open model                |

#### Doubao

| Model              | Capability | Price  | Description                        |
| :----------------- | :--------- | :----- | :--------------------------------- |
| **Seed 2.1 Pro**   | Quality    | High   | Flagship Chinese reasoning         |
| **Seed Evolving**  | Quality    | High   | Balanced Chinese for everyday work |
| **Seed 2.1 Turbo** | Balanced   | Medium | Fast Chinese for quick tasks       |
| **Seed 2.0 Mini**  | Fast       | Low    | Fast Chinese for simple queries    |
| **Seed 2.0 Pro**   | Quality    | Medium | Strong Chinese for demanding tasks |
| **Seed 2.0 Lite**  | Fast       | Low    | Fast Chinese for everyday chat     |

#### Qwen

| Model         | Capability | Price | Description                 |
| :------------ | :--------- | :---- | :-------------------------- |
| **Qwen3 32B** | Balanced   | Low   | High-performance open model |
| **Qwen3 14B** | Fast       | Low   | Lightweight open model      |

#### DeepSeek

| Model                 | Capability | Price  | Description                      |
| :-------------------- | :--------- | :----- | :------------------------------- |
| **DeepSeek V4 Pro**   | Quality    | Medium | Premium reasoning model          |
| **DeepSeek V3.2**     | Quality    | Low    | Cost-effective for everyday chat |
| **DeepSeek V4 Flash** | Balanced   | Low    | Fast, low-latency processing     |

#### Zhipu

| Model       | Capability | Price  | Description                            |
| :---------- | :--------- | :----- | :------------------------------------- |
| **GLM-5.2** | Quality    | Medium | Flagship open-source agentic reasoning |
| **GLM-5**   | Quality    | Medium | Open-source bilingual all-rounder      |
| **GLM-4.7** | Balanced   | Medium | Reliable reasoning for everyday work   |

***

## 🔑 Bring Your Own Model

Add custom models hosted on any OpenAI-compatible API endpoint using your own API credentials.

<Frame caption="Bring your own model integration panel for custom API endpoints.">
  <img src="https://mintcdn.com/memverge-1b064d4d/9UyDf9zIDz_f9H7C/configuration/assets/desktop-models-byom.png?fit=max&auto=format&n=9UyDf9zIDz_f9H7C&q=85&s=be28852567bb730e54ce5edca5aa5467" alt="MemoryBox Bring Your Own Model panel" width="900" height="350" data-path="configuration/assets/desktop-models-byom.png" />
</Frame>

### Configuration Fields

* **Display name:** The custom label that will appear in your chat model picker selection list.
* **Base URL (`https://...`):** The network endpoint target hosting your API service.
* **Upstream model id:** The exact model string identifier expected by your target provider engine.
* **API key:** Your private API credential token.

Click **Add model** to register the endpoint into your active available model suite. Registered custom models will appear in a list directly below the form.