Cherry Studio

Windows

INFO

This guide is based on the Cherry Studio v1.5.6 interface; other versions may differ slightly.

Preparation

• A valid OpenRouter API Key
• Cherry Studio v1.5.6 installed on Windows

⬇️Cherry Studio v1.5.6 for Windows

Installation & First Launch

1. Complete the installation and start Cherry Studio.

2. Go to the Settings panel on the left sidebar, and you’ll see several key sections:
   • Model Services: configure all large language models and providers.
   • Global Memory: helps the assistant understand you better in future chats (optional).
   • Other Settings → Web Search: it’s recommended to switch the search provider to Bing for fresher results.
   • Default Models: used when no model is manually selected in a conversation.

Add OpenRouter in “Model Services”

1. Open Settings → Model Services.
2. Find OpenRouter in the provider list on the left.
3. Enter the API Base URL (usually https://openrouter.ai/api/v1/) and your API Key.
4. Click Add Model and add the following models one by one (use the same text for both “Model ID” and “Model Name”).

Recommended model IDs (copy and add one by one)：

openai/gpt-5

openai/gpt-5-nano

google/gemini-2.5-pro

google/gemini-2.5-flash

::: Tips Leave Group Name empty; group later if needed.

If testing fails, first check: whether the API Key is valid, the Base URL is /api/v1/, and whether the network is being blocked. :::

Set “Default Models”

Open Settings → Default Models and choose as follows:

Default assistant models:

openai/gpt-5

google/gemini-2.5-pro

Topic naming model:

openai/gpt-5-nano

Translation model: optional; if needed choose

openai/gpt-5

Recommendations

gpt-5：strong general capability and stable alignment; suitable for daily use and creation.

gemini-2.5-pro: good at reasoning/tool use; often better price/speed.

gpt-5-nano: cost-effective for lightweight tasks (e.g., auto naming, summaries).

Start Chatting & Switch Models

Return to Assistants on the left and select “Default Assistant” to start chatting. You can switch models at the top of the chat page. Common choices:

openai/gpt-5

google/gemini-2.5-pro

Advanced: Reasoning Mode & Web Search

5.1 Reasoning Mode (Chain-of-thought length)

Click the 💡 icon at the bottom-left of the input box to open reasoning settings:

Default: the model automatically controls reasoning length based on task complexity (recommended).

In general, longer reasoning yields higher answer quality but also higher latency.

5.2 Web Search (Bing)

When you need time-sensitive information, enable Web Search and choose Bing. When enabled, the current message will automatically call the web tool to assist with the answer.

FAQ

Q1:Can’t see the models in the list after adding?

A：Make sure you clicked Add Model under the OpenRouter provider for each entry and saved.

Q2:Answer quality fluctuates a lot?

A：Turn on Reasoning Mode; for complex tasks, prefer openai/gpt-5 or gemini-2.5-pro; enable Bing Web Search when needed.