Who wins when a tech giant opens its gates? Developers get power; platforms get competition. I see this shift in how we build software.

Google’s Generosity Shines: Open-Source, Free Coding Agent Supports Million-Token Context, Multimodal Input, and MCP

Open Source and Free! Google Targets Programming Agents.

I read the announcement with interest as Google launched Gemini CLI, allowing you to access Gemini directly from your terminal. They claim this offers the “highest free tier in the industry”:

• Gemini 2.5 Pro with a 1 million context window
• Up to 60 model requests per minute
• A daily cap of 1,000 requests

The official announcement also emphasized a key point that I found particularly noteworthy:

While Gemini CLI is powerful for coding, it’s not limited to programming tasks.

You can invoke Veo and Imagen to generate videos directly from the command line—it takes just one sentence:

I think free tiers lower barriers for indie developers and small teams. For creators, open-source tools reduce dependency on proprietary vendor lock-in.

The news sparked immediate discussion, with the GitHub repository stars skyrocketing to 10.8k overnight.

Some netizens even declared RIP for Cursor.

For more direct competitors like Claude Code and OpenAI’s Codex CLI, the response was simple: “Free” trumps all.

In short, the tech giant Google is showing off its deep pockets with a significant investment.

Using Gemini in the Command Line

Let’s look at the details of Gemini CLI.

According to the official introduction, its capabilities cover code understanding, file operations, command execution, and dynamic troubleshooting.

Simply put, with Gemini CLI, users can use natural language in the command line to instruct the Gemini model to write code and debug.

More specific application examples include coding features:

• Supports a 1 million context window, enabling querying and editing of large codebases.
• Integrates with GitHub to support automated operational tasks, such as querying pull requests and managing code branches.

Tool calling capabilities include:

• Multimodal support allows generating apps directly from PDFs or sketches.
• Supports MCP (Model Context Protocol), enabling image and video generation using Imagen, Veo, Lyria, etc.
• Allows the use of Google Search.

User tests also show Gemini CLI’s performance in parallel processing:

Additionally, Gemini CLI shares technology with Google’s AI coding assistant, Gemini Code Assist.

Regarding the open-source license, Gemini CLI follows the Apache 2.0 license.

bash npx https://github.com/google-gemini/gemini-cli