How to Connect Claude API in Chatbox & Cherry Studio (2026 Tested)

Not everyone wants to write API code — sometimes you just want a solid desktop client where you can chat with Claude, draft documents, refactor code, and analyze files.

Chatbox and Cherry Studio are two of the most popular AI desktop clients that both support custom API endpoints. This guide walks you through connecting them to Claude’s full model lineup via ClaudeAPI.com — accessible from anywhere, no VPN required, and ready in under 3 minutes.

📌 All data in this article was tested in April 2026 using the OpenAI-compatible endpoint at ClaudeAPI.com — direct connection, no proxy needed.

Step 0: Get Your Claude API Key

Before configuring any client, you’ll need a working API key.

Get one at ClaudeAPI.com in 3 steps:

1.Create an account — Sign up with your email at ClaudeAPI.com

2.Top up your balance — Multiple payment methods supported; new accounts receive a $5 free credit to get started

3.Generate your key — Go to Dashboard → Token Management → Add New Token → Copy & save it

💡 For a detailed walkthrough, see：Claude API Pricing & Model Selection Guide (2026)

Once you have your key, note down these connection details:

💡 Both Chatbox and Cherry Studio connect via the OpenAI-compatible format. Just set the Base URL to https://claudeapi.com — direct access, no proxy required.

Claude Model Benchmarks (All 5 Models Verified)

Before diving into client setup, we ran real tests across all available Claude models using the OpenAI-compatible endpoint — so you know exactly what speed and token usage to expect.

Test Conditions

• Test Date：2026
• Network：Direct connection, no VPN
• API Format：OpenAI-compatible (claudeapi.com)
• SDK：Python OpenAI SDK
• Test Prompt："Briefly introduce the advantages of Python"（max_tokens=200）

Speed & Token Usage — 5 Models Benchmarked

💡 Haiku 4.5 is the fastest (4.03s) — ideal for simple tasks and high-volume batch processing. Sonnet 4.6 offers the best price-to-performance ratio for everyday use. For detailed pricing, see the Claude API Pricing & Model Selection Guide (2026)。

Streaming Output Test

Tested with stream=True ：

Streaming works perfectly — in Chatbox and Cherry Studio, you’ll see the response appear word by word in real time.

Multi-Turn Conversation Test

Tested a 3-turn conversation with a System Prompt:

Total time for 3 turns: 22.41s. Context was maintained throughout — responses were accurate and consistent.

Long-Form Generation Test

Connecting Chatbox to Claude API

What is Chatbox?

[ChatBox]is a free, open-source AI desktop client available on Windows, macOS, Linux, iOS, Android, and Web. Key features include:

• 🗨️ Multi-model chat (GPT, Claude, Gemini, and more)
• 📄 Document & image conversations (PDF, Word, code files)
• 💻 Code highlighting and preview
• 📝 Markdown / LaTeX rendering
• 🔒 Local data storage — your conversations stay on your device
• 🔌 MCP service integration

Download & Install Chatbox

Head to chatboxai.app and grab the installer for your platform:

Configuring Claude API in Chatbox — Step by Step

Step 1: Open Settings

Launch Chatbox → click the ⚙️ Settingsicon in the bottom-left corner.

Step 2: Select a Model Provider

In the Settings panel, find “AI Model Provider” and choose:

Add Model Provider (OpenAI API Compatible)

Step 3: Fill In Your API Configuration

Step 4: Add Your Model

Click “+ Add” to add at least one model.

You can optionally check the capability flags for that model (vision, function calling, etc.) — if left unchecked, it defaults to text-only mode.

Step 5: Save and Test

Click Save, go back to the chat window, and send a message.

Try this prompt:

Hi, introduce yourself in one sentence.

Expected response:

“Hi there! Great to meet you! What can I help you with today? 😊”

— claudeapi (claude-sonnet-4-6) · 80 tokens (actual test result)

Switching Claude Models in Chatbox

You can switch models anytime from the chat interface or settings — just update the model name:

💡 Quick model-picking guide:

• Everyday chat → Sonnet 4.6
• Deep analysis & complex reasoning → Opus 4.6
• Simple tasks where speed matters → Haiku 4.5

Connecting Cherry Studio to Claude API

What is Cherry Studio?

[Cherry Studio] is an open-source AI desktop client for Windows, macOS, and Linux. It stands out with:

• 🤖 300+ AI models supported, compatible with 50+ AI providers

• 🧩 300+ built-in AI assistants ready to use

• 📚 Knowledge base, AI image generation, and translation tools

• 🖥️ Local model support via Ollama and LM Studio

• 🆓 Completely free and open-source

Download Cherry Studio

Grab the installer for your platform at cherry-ai.com/download .

Configuring Claude API in Cherry Studio — Step by Step

Step 1: Open Settings

Launch Cherry Studio → click the ⚙️ Settings icon in the left sidebar.

Step 2: Add a Custom Provider

In the Settings panel, navigate to “Model Services” or “Provider Management”:

Click “Add Provider” Select OpenAI or Anthropic as the provider type Set the provider name to something like ClaudeAPI **Step 3: Fill In Your API Configuration

Step 4: Add Your Models

Once the provider is configured, manually add the models you want to use:

Under the ClaudeAPI provider, click “Add Model” and enter the model IDs one by one:

• claude-opus-4-6
• claude-haiku-4-5-20251001
• claude-sonnet-4-5-20250929
• claude-sonnet-4-20250514

Step 5: Start Chatting

1. Go back to the chat interface

2.Switch to the ClaudeAPI provider in the model selector

3.Select claude-sonnet-4-6

4.Send a test message

💡 Pro tip: Cherry Studio lets you switch between models mid-conversation — great for comparing response quality on the fly.

Want to see more advanced Cherry Studio workflows? Check out our guide: How to Use Claude API in Cherry Studio for SEO — A Complete Beginner’s Guide。

Chatbox vs. Cherry Studio: Which One Should You Pick?

💡 TL;DR: Need it on your phone too? → Go with Chatbox. Want 300+ built-in AI assistants and a feature-packed desktop experience? → Go with Cherry Studio. Both are free — there’s no harm in trying both.

Why ClaudeAPI.com Instead of Anthropic Direct?

ClaudeAPI.com is the fastest way to get Claude API access without the usual friction — no geo-restrictions, no international card required.

Want to go deeper? Check out our guide:考 How to Use Claude API for SEO in 2026: 7 Workflows + Ready-to-Use Prompts。

Pro Tips

Customizing the System Prompt

Set a custom system prompt in your conversation settings to give Claude a specific role. Here’s one we use in production:

You are a senior Python engineer with deep expertise in code review 
and system architecture. Keep your answers concise.

You are a senior Python engineer with deep expertise in code review 
and system architecture. Keep your answers concise.

Real-world result: Across 3 rounds of back-and-forth, responses were sharp, accurate, and to the point — total time: 22.41s.

Keeping Costs Under Control: Model Selection Guide

Based on real usage data, here’s what we recommend for each use case:

💡 Real benchmark: We used Sonnet 4.6 to generate a 478-word article (“Why Every Developer Should Learn AI”) — 35 tokens in, 476 tokens out, completed in 42.22s. Total cost at ClaudeAPI.com pricing: ~$0.003.

Document Chat

Both Chatbox and Cherry Studio support uploading documents and chatting with Claude directly:

• 📄 Upload a PDF → Ask Claude to summarize key points or answer questions

• 💻 Upload a code file → Let Claude do a code review

• 🖼️ Upload a screenshot → Have Claude analyze UI issues or error messages

• 💡 Document chat leverages Claude’s multimodal capabilities. For best results, use Sonnet 4.6 or Opus 4.6.

FAQ

Messages not going through after setup / getting errors?

Run through this checklist:

• [ ] Is the Base URL set to https://claudeapi.com
• [ ] Did you copy the full API Key (starts with sk-)?
• [ ] Is the model name spelled correctly?
• [ ] Does your ClaudeAPI.com account have sufficient balance?

Getting “model not found” or “invalid model”?

This is almost always a model name typo. Model IDs are case-sensitive. Here are the verified, working model IDs:

claude-opus-4-6
claude-sonnet-4-6
claude-haiku-4-5-20251001
claude-sonnet-4-5-20250929
claude-sonnet-4-20250514

claude-opus-4-6
claude-sonnet-4-6
claude-haiku-4-5-20251001
claude-sonnet-4-5-20250929
claude-sonnet-4-20250514

Claude API responses feel slow — is that normal?

Here are real latency benchmarks for a simple prompt with ~200 tokens of output:

If you’re seeing significantly slower times, check your network connection. Long-form generation (500+ tokens) typically takes 30–50s — that’s expected behavior.

Can’t see any models in Cherry Studio?

After adding a provider, you need to add models manually. In the provider settings, click “Add Model” and type in the model ID (e.g., claude-sonnet-4-6).

Does it support streaming output?

Yes. Streaming has been tested and works correctly:

• Time to first token: 5.38s (Sonnet 4.6)

• Response streams in progressively, delivered across 13 chunks

Both Chatbox and Cherry Studio have streaming enabled by default.

Is it complicated to connect Chatbox / Cherry Studio to Claude API?

Not at all — you can be up and running in under 3 minutes. ClaudeAPI.com provides a globally accessible endpoint that works from anywhere without a VPN. Just plug in the base URL and your API key, and you’re good to go.

Quick Test: Verify Your API Key with Code

Want to confirm your key works before setting up a GUI client? Run this Python snippet:

from openai import OpenAI

client = OpenAI(
    api_key="sk-your-Key",
    base_url="https://claudeapi.com"
)

# Basic test
response = client.chat.completions.create(
    model="claude-sonnet-4-6",
    messages=[{"role": "user", "content": "Hello, introduce yourself in one sentence."}],
    max_tokens=100
)
print(response.choices[0].message.content)

# Streaming test
stream = client.chat.completions.create(
    model="claude-sonnet-4-6",
    messages=[{"role": "user", "content": "What are the strengths of Python?"}],
    max_tokens=200,
    stream=True
)
for chunk in stream:
    if chunk.choices and chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)

from openai import OpenAI

client = OpenAI(
    api_key="sk-your-Key",
    base_url="https://claudeapi.com"
)

# Basic test
response = client.chat.completions.create(
    model="claude-sonnet-4-6",
    messages=[{"role": "user", "content": "Hello, introduce yourself in one sentence."}],
    max_tokens=100
)
print(response.choices[0].message.content)

# Streaming test
stream = client.chat.completions.create(
    model="claude-sonnet-4-6",
    messages=[{"role": "user", "content": "What are the strengths of Python?"}],
    max_tokens=200,
    stream=True
)
for chunk in stream:
    if chunk.choices and chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)

Wrap-Up: Claude API in 3 Minutes

No coding required. Works from anywhere. Here’s all it takes to get Claude running in a desktop client:

Quick-Reference Config

Base URL:  https://claudeapi.com
API Key:   sk-...
Models:   claude-sonnet-4-6（everyday use） / claude-haiku-4-5-20251001（fast + budget-friendly）

Base URL:  https://claudeapi.com
API Key:   sk-...
Models:   claude-sonnet-4-6（everyday use） / claude-haiku-4-5-20251001（fast + budget-friendly）

What Our Testing Confirmed

• ✅ All 5 Claude models passed — working correctly
• ✅ Streaming output works — time to first token: 5.38s
• ✅ Multi-turn context maintained across 3 rounds — 22.41s total
• ✅ Long-form generation works — 478-word article in 42.22s

👉 ** Get your API key at claudeapi.com and put Claude to work as your AI assistant today.**

Written by the ClaudeAPI.com engineering team. All benchmarks were collected in 2026. Have questions or running into issues? Visit ClaudeAPI.com — we’re happy to help.