Ollama Model Interaction

Ollama provides multiple ways to interact with models, the most common being inference operations via the command line.

1. Command-Line Interaction

Interacting with the model directly via the command line is the simplest way.

Run the Model

Useollama runcommand to start the model and enter interactive mode:

ollama run <model-name>

For example, we download the qwen3.5 model:

Example

ollama run qwen3.5

After startup, you can directly input questions or commands, and the model will generate responses in real time.

>>> 你好，你能帮我写一段代码吗？
当然可以。但是首先我想知道您希望在哪种编程语言中实现这个功能（例如Python、JavaScript等）和要解决什么问题或者完成的任务是什么样的例子呢？这样我们可以为您提供更准确的
内容，同时也方便我帮助你写出最适合您的代码片段。


>>> 写一段 python hello world
当然可以！这是一个简单的 "Hello, World!" 程序：

```python
print("Hello, World!")
```
这个脚本会输出 `Hello, World!`，并将其打印到控制台上。这只是最基本的 Python Hello world示例；Python是一种解释型、通用型的编程语言以其简洁性和易读性而闻名。它还允许
用户在代码中插入变量和表达式来创建复杂的行为。

Exit Interactive Mode

In interactive mode, type/byeor pressCtrl+d to exit.

2. Single-Command Interaction

If you only need the model to generate a single response, you can pass the input directly on the command line.

Using Pipe Input

Pass the input to the model via a pipe:

Example

echo "Who are you?" | ollama run qwen3.5

The output is as follows:

Using Command-Line Arguments

Pass the input directly on the command line:

ollama run qwen3.5 "Python 的 hello world 代码？"

The output is as follows:

在 Python 中，"Hello World!" 通常是这段简单的脚本：
```python
print("Hello World!")
```
当你运行这个程序时，它会输出 `Hello, World`。这是因为 print() 函数将字符串 "Hello, World" 打印到标准输出设备 (stdout) - 也就是你的屏幕上显示的信息（在这种情况下是命
令行终端或类似的工具中运行 Python 脚本时，它会直接写入控制台。

3. Multi-Turn Conversation

Ollama supports multi-turn conversations, and the model can remember context.

Example

>>>Hello, can you help me write a piece of Python code?
Of course! Please tell me what functionality you need to implement.

>>>I want to write a function to calculate the Fibonacci sequence.
Okay, here is a simple Python function:
def fibonacci(n):
    if n <= 1:
        return n
    else:
        return fibonacci(n-1) + fibonacci(n-2)

4. File Input

You can pass file content as input to the model.

Assume the content of input.txt is:

 Python 的 hello world 代码？

Use the content of input.txt as input:

ollama run qwen3.5 < input.txt

5. Custom Prompt

Define custom prompts or system instructions through the Modelfile to make the model follow specific rules during interactions.

Create a Custom Model

Write a Modelfile:

Example

FROM qwen3.5
SYSTEM "You are a programming assistant, dedicated to helping users write code."
Then create the custom model:
ollama create example-coder -f ./Modelfile

Run the custom model:

ollama run example-coder

6. Interaction Logs

Ollama records interaction logs for convenient debugging and analysis.

View logs:

ollama logs
Other Extensions