ChatBox Mobile Installation & Configuration Guide: Complete Setup for Claude Model Access via API

ChatBox is a widely popular cross-platform AI chat client that supports connecting to various large language models on mobile devices through third-party APIs. Unlike official web interfaces (such as claude.ai), ChatBox allows users to input their own API keys to call models, bypassing regional restrictions or subscription plans of official clients. The core value of such tools lies in "one client, multiple models" — users can switch between OpenAI GPT, Anthropic Claude, Google Gemini, and other providers' models within a single interface, significantly lowering the barrier to entry. ChatBox supports Windows, macOS, Linux, Android, and iOS across all platforms, and conversation records are stored locally, balancing both privacy and convenience. This article provides a detailed guide on how to install ChatBox on your phone and configure the API to access models like Claude Haiku 4.5 and Claude Opus 4.5.

Prerequisites

Before starting the configuration, you'll need to prepare the following:

• ChatBox Client: Download the ChatBox app for your phone from official channels (supports Android and iOS)
• API Key: Obtain a valid API Key from a third-party API service provider
• API Address: This tutorial uses the API host address 200.xstx.info

This tutorial supports models including claude-haiku-4-5-20251001 and claude-opus-4-5-20251101-thinking, among five major models covering multiple versions from lightweight to flagship in the Claude series. Anthropic's Claude model family is divided into three tiers by capability: Haiku (lightweight, focused on speed and low cost), Sonnet (mid-tier, balancing performance and price), and Opus (flagship, strongest reasoning capability). The numeric codes in model IDs contain version and release date information — for example, claude-haiku-4-5-20251001 indicates the Haiku version of the Claude 4.5 series, a snapshot released on October 1, 2025. The thinking suffix in claude-opus-4-5-20251101-thinking indicates that the model has Extended Thinking enabled — the model performs a step-by-step internal reasoning process (similar to Chain-of-Thought) before generating the final answer, which significantly improves performance on complex tasks such as mathematical proofs, code debugging, and logical analysis compared to the standard version.

Step 1: Access Settings to Configure the Model Provider

Open the ChatBox app on your phone and follow these steps:

1. Tap the hamburger menu icon in the upper left corner
2. Select "Settings" from the sidebar that appears
3. Find and tap the "Model Provider" option

On the model provider configuration page, fill in the following key information:

• Name: Enter 200XSTX.INFO (for easy identification later)
• API Mode: Select Cloud API Compatible mode

Selecting Cloud API Compatible mode is crucial — it's the key setting that ensures third-party APIs work properly. This mode essentially follows the Chat Completions API interface specification originally defined by OpenAI (using /v1/chat/completions as the endpoint, Bearer Token authentication, and passing a messages array in JSON format as the standard protocol). Since OpenAI was the first to open its API at scale for large language models, its interface format has become the de facto industry standard. Currently, the vast majority of third-party API relay services and open-source model inference frameworks (such as vLLM and Ollama) are compatible with this specification, so selecting this mode ensures ChatBox seamlessly connects with virtually all relay services.

Step 2: Enter the API Key and Host Address

After completing the basic setup, configure the API authentication information:

1. Tap the "Add" button
2. Copy and paste the API key you received upon purchase into the corresponding input field
3. In the API Host field, enter: 200.xstx.info

Make sure the key and host address are entered correctly — any extra spaces or characters may cause connection failures. It's recommended to copy and paste directly to avoid manual input errors.

The third-party API relay service here acts as a proxy layer between the user and the original model provider. Users send requests to the relay server (such as 200.xstx.info in this article), which then forwards the requests to Anthropic's official API endpoint, retrieves the response, and returns it to the user. The advantages of this architecture include: first, it solves the problem of certain regions being unable to directly access official APIs; second, relay service providers typically support pay-per-use billing, so users don't need to apply for official API accounts or bind international credit cards; third, multiple model providers' services can be aggregated under a single unified key. Note that using a relay service means your conversation data passes through third-party servers, so you should assess security risks when handling sensitive information.

Step 3: Create a New Model and Configure Parameters

Once the API connection information is configured, you need to manually add specific models:

1. Tap the "New" button
2. In the Model ID field, enter the model name you want to use (e.g., claude-haiku-4-5-20251001)
3. Confirm the API address is: https://200.xstx.info/v1
4. Context Window: Enter 200000 (i.e., 200,000 tokens)
5. Max Output: Enter 5000 tokens

Parameter Explanation

• Context Window of 200,000 tokens: The context window refers to the maximum amount of text the model can "see" and process in a single conversation, measured in tokens. A token is the basic unit of text processing for large language models and is not simply equivalent to a single character or word: for English, 1 token corresponds to approximately 4 characters or 0.75 words; for Chinese, 1 Chinese character typically consumes 1-2 tokens. A 200,000-token context window means you can input approximately 100,000-150,000 Chinese characters in a single conversation (equivalent to a medium-length book), which is extremely useful for scenarios like long document summarization, contract review, and codebase analysis. Note that the context window includes the total of both input and output, so the actual available input space needs to be reduced by the maximum output token count.

• Max Output of 5,000 tokens: The Max Tokens parameter determines the upper limit of the model's single response length. Setting it to 5,000 tokens can generate approximately 3,000-4,000 Chinese characters, roughly equivalent to a medium-length article. If the model's response naturally concludes before reaching this limit, it won't be forcibly padded; however, if the response content exceeds this limit, the output will be truncated. For scenarios requiring long-form content generation (such as complete technical documentation or lengthy translations), you can increase this parameter to 8,000 or even higher, but note that longer outputs mean higher API call costs and longer wait times. The maximum output limit for Claude 4.5 series models is typically 8,192 or 16,384 tokens, depending on the model version.

If you need to use other models, repeat the above creation steps to add them one by one. All five supported models can be configured in the same way.

Step 4: Test the Connection and Save

After all parameters are filled in, the most critical step is verifying the configuration:

1. Tap the "Test" button on the page
2. Wait for the system to send a test request
3. If it shows Test Successful, the configuration is correct
4. Tap "Save" to complete all settings

If the test fails, check the following items one by one:

• Whether the API key is correct and hasn't expired
• Whether there are spelling errors in the API host address
• Whether the API mode is set to "Cloud API Compatible"
• Whether the network connection is working properly

Getting Started

After saving the configuration and returning to the ChatBox home page, you'll see the Claude models you just added in the model selection list. Select any model to start a conversation.

Usage Recommendations

• Daily conversations and lightweight tasks: Claude Haiku 4.5 is recommended for its fast response speed and low cost. As a lightweight model, Haiku excels at simple Q&A, text rewriting, everyday translation, and similar tasks, with API call costs at only a fraction of Opus — making it ideal for high-frequency use scenarios.
• Complex reasoning and deep analysis: Claude Opus 4.5 Thinking version is recommended for its stronger chain-of-thought reasoning capabilities. This version performs internal step-by-step reasoning (Extended Thinking) before answering, and although response times are longer, accuracy is significantly improved for mathematical problem solving, multi-step logical reasoning, and complex code generation.
• Long document processing: Take full advantage of the 200,000-token context window to input complete documents for analysis and summarization in one go. For example, you can paste an entire legal contract, an academic paper, or even all the source code from a small code repository into the conversation, allowing the model to perform holistic understanding and analysis — avoiding context loss caused by segmented input.

The entire configuration process takes only about 3-5 minutes, after which you can use powerful Claude AI models anytime, anywhere on your phone.