Did it flop in less than a day after its release? Google's free AI programming product isn't just another "rebranded" Claude.

Bringing the latest and most powerful model to the hottest track, this is the most apt description of Google's launch of Gemini CLI yesterday.

The field of AI programming, even if limited to the scope of programming agents, is now extremely crowded. As soon as the CLI was released, netizens couldn't wait to gather all similar products together for a head - to - head comparison.

However, within less than 20 hours of its release, various "failures" followed one after another.

The main selling points of Gemini CLI are as follows:

• Fully open - source + free: Licensed under Apache 2.0, with 1,000 free requests per day.
• Multi - task framework: It not only supports programming but also command - line tasks such as file management, content generation, and script control.
• Lightweight + scriptable: You can embed it in your terminal scripts or use it as an automated agent.

However, soon after its release, the login and authentication processes were not as smooth as expected.

Especially in China, there were many restrictions on logging in. Moreover, even if you managed to get it running, there were all kinds of bugs.

Some netizens mentioned that getting it to write code is still unstable (it requires a bit of luck). However, it can be useful in code interpretation, verification, and troubleshooting.

According to the feedback from netizens who participated in the internal test, it indeed requires some well - designed usage methods.

The most hilarious part is that after netizens' overnight testing, it was found that this is actually a "wrapper" - it wraps a competitor's product... It's just a repackaging of Claude Code's CLI code.

Embarrassing... Sweating bullets, bro...

What is CLI

Google's entry into the AI programming arena with CLI is sufficient proof of the importance of this field.

Gemin CLI is similar (identical) to Claude Code. Both are practical tools that allow ordinary people to use the command line to have AI assist with daily tasks. Compared to programming tools, it is more like an open - source "AI agent" - you can use natural language in the terminal to call the Gemini 2.5 Pro model to handle various tasks. It's not limited to writing code; it can also explain documents, search the web, generate content, and even control local tools.

From some use cases found online, we can see that the use of CLI is not just for programming; programming is just one of its means.

For example, a Japanese netizen deployed a weather forecast application through CLI and could directly push the day's weather information to LINE.

Others have created an application for organizing invoices.

It can also be used to create an official website for a coffee shop and to extract and integrate information from PDFs.

These are all very common usage scenarios. It can be seen that unlike Cursor, CLI takes the route of a multi - tasking "assistant" rather than a dedicated code - collaboration tool.

Of course, it can also handle more complex tasks. For example, an engineer from the Google team posted a use case of creating a 3D animation with CLI.

And the official released a video created by calling the Veo3 video - generation model: A cat's adventure around the world.

Command Line, the Starting Point of Vibe Coding

So, the question is: What is the command line?

The command line (Command Line) is a form of human - computer interaction. Instead of clicking icons or using the mouse, you directly type commands in a black box, and the system returns the results.

Generally speaking, the command line was once considered a tool only used by developers - a black background, typing commands, and no need for a mouse, and even more so by intermediate and advanced developers. For example, under the design of Gemini CLI, the command line has become a lightweight and flexible entry for an AI assistant: You just need to open the terminal and type a natural - language request, such as "Explain this code" or "Check the content of this file", and the AI will return the results to you.

This is different from many AI programming tools that people are familiar with today (such as Cursor). Cursor is a graphical - interface tool that helps programmers write code and polish comments. It is an "assistant during coding".

On the other hand, Gemini CLI and Claude Code are "task agents". You don't necessarily have to write code. It can handle text, files, web pages, and data, and can even be embedded in existing script processes. You just type a question, and Gemini 2.5 Pro will answer you. It's actually a form of chatbot conversation, but it takes place in the terminal rather than on a web page.

This means that ordinary content creators can also use the terminal - style interaction to experience the "prompt + execution + feedback + refinement" cycle with a low threshold.

In other words, what Google has created this time is not a programmer's tool like Cursor. Instead, it takes the form of an agent, allowing you to control your workflow as if you were having a natural - language conversation with a colleague. For those who are not full - time programmers but need to process a large amount of information and produce content, it means that you can finally use a conversational approach to handle some tedious tasks that previously had to be done manually.

So, did Google's CLI copy Claude Code?

Actually, there are some differences. Both Gemini CLI and Claude Code are command - line AI programming assistants, but the former is more flexible, open - source, and applicable to multiple scenarios with a wider range of applications; the latter is more focused, mature, and charges a fee.

People who need in - depth project collaboration may prefer the professionalism of Claude Code. It's more like a programming partner with "memory + execution" capabilities, capable of delving into your project, running tests, generating pull requests, and automating more deeply.

For users who want to conduct simple experiments and need multi - task assistance, Gemini CLI is a no - threshold starting point. It provides you with an entry to summon AI at any time. You don't need to write code; you can just give a command to let the AI "do" things, whether it's writing text, checking files, or running commands.

For Google, although the CLI is not perfect, has many bugs, and its future is uncertain, its release is a statement. AI is no longer just a chat toy on the web page but the default entry integrated into daily work processes.

For creators or developers who neither want to write a full set of code nor are satisfied with pure chat - based interaction, getting started with Gemini CLI will be a bit easier. Looking on the bright side, you can finally use your native language to control an AI that "does the work for you".

To be honest, vibe coding is a practice that is hard to define but actually exists. And Gemini CLI may be the starting point for it to become more concrete and accessible.

This article is from the WeChat official account "APPSO". Author: Discovering tomorrow's products. Republished by 36Kr with permission.