The free era of Copilot ends! The official version is 67 yuan/month, and students and maintainers of popular open source projects can get it for free

Mengchen sent from Aofei Temple
Quantum Bit | Public Account QbitAI

GitHub Copilot , the coding tool and AI assistant for programmers , has ended its internal testing and is now open to everyone!

The official version is priced at $10/month (about RMB 67) , and the annual payment is discounted to $100/year (about RMB 669)

From now on, you can apply for a 2-month free trial without waiting in line. It is also free for certified students and maintainers of popular open source projects .

When I saw this I bought two copies on the spot.

I tried UnionPay + Visa dual-currency credit card and domestic bank debit card through Paypal , and both were successful.

After a successful purchase, you can install and use your favorite tool. Currently, only four code editors and IDEs are supported: VS Code, Visual Studio, JetBrains full package, and Neovim.

△ More versions will be released in the future

The official version of Copilot supports multiple mainstream programming languages, and performs particularly well in Python, JavaScript, TypeScript, Ruby, and Go .

In terms of usage, you can write a comment to let AI generate the entire function, or you can write half of the code and let AI automatically complete it based on the context.

GitHub believes that this can help programmers save time and focus on solving more serious problems.

GitHub also hopes that Copilot can help people master a new programming language.

With an AI assistant to help you fix bugs or save time searching for documents, people can be more confident when exploring unknown areas.

So, how useful is this AI tool that aims to be a good partner for programmers?

See what beta testers say

According to GitHub official statistics, Copilot has 1.2 million users in the past year of open testing.

Across all files Copilot has participated in, nearly 40% of the code was written by AI in popular languages ​​such as Python, and this proportion is expected to rise.

In terms of internal beta user feedback, many people both at home and abroad said that "after using it for a year, I can't live without it . "

In terms of user experience, some netizens believe that the way they write code has been changed by Copilot. Specifically, there are three points:

1. Part of my job has shifted from writing to planning. As a human, I can observe and fix some code without having to do everything myself.

2. I have a higher tolerance for redundant code. Letting AI do repetitive work and writing the code in more detail can improve readability.

3. I am more willing to refactor the code. For those codes that are already usable but not well written, Copilot can flexibly complete the refactoring, such as splitting complex functions or abstracting key parts.

In addition, some internal test users said that when using Copilot, they don’t have to repeatedly look up different syntax when frequently switching programming languages, which can save a lot of time.

Now that the free internal beta has ended, is the official version worth the $10 per month price?

Many users who have experienced the beta version said:

Although Copilot has some shortcomings, it is definitely worth it according to the income level of programmers !

△ Add emoticons to him

Some domestic users also said that with such a cool function, they expected the price to be more expensive, but now it seems very cheap.

However, whether Copilot should charge a fee has become a focus of controversy.

In the VS Code plug-in market, Copilot has always had 5-star reviews. Since the announcement of charging today, many 1-star negative reviews have poured in, bringing the overall score to 4.5.

The reason is that some people are disgusted by the business model of first providing free services to make people addicted and then charging for them.

Another point that has been criticized is that Copilot should not charge developers for using the code they write for training.

The idea that "everyone who has submitted code to GitHub should be able to use it for free" has gained a lot of recognition.

However, this is just one of the many controversies Copilot has faced since its launch.

Controversy continues

Less than two weeks after Copilot was released, it was criticized by many people in the open source community.

The core problem is that the official admitted that all public codes on GitHub would be used to train Copilot, and no distinction would be made between license categories .

For many licenses that are not allowed to be modified for commercial use, Copilot will face infringement issues now that it charges a fee.

GitHub has currently proposed a temporary solution to this problem.

After successfully purchasing Copilot, two settings will appear: Do you allow Copilot to use public code snippets? And do you allow Copilot to use your code snippets as training data?

Whether this can circumvent copyright issues remains to be verified.

GitHub is determined to promote the commercialization of Copilot and plans to launch an enterprise version within this year. Copyright issues will definitely be resolved by then.

In addition to copyright issues, the academic community has also done a lot of research on Copilot and the OpenAI Codex behind it.

In one of the papers, researchers used Copilot to generate a total of 1,692 programs for 89 different scenarios and found that 40% of the programs had bugs or security vulnerabilities.

Faced with these problems, some netizens feel that they are still not comfortable using Copilot for formal work, and some companies currently prohibit employees from using Copilot.

Many people also said that they would mainly use Copilot for developing amateur projects and students for learning programming.

In this case, don’t forget that Copilot is also free.

Free prostitution strategy

The official version of Copilot promises to be free for certified students and maintainers of popular open source projects .

For student certification , the process is the same as before for free private libraries, which is to apply for GitHub Student Developer Pack.

Students aged 13 or above who are enrolled in junior high schools, high schools, colleges and other educational institutions that can award degrees or diplomas can apply.

You need to have a verifiable education email address and upload your identity proof.

Domestic schools have been tested by netizens and have the opportunity to pass the certification by uploading student ID cards or the Ministry of Education’s online academic certification report from the China Higher Education Student Information and Career Center. The review period is approximately 2-3 weeks.

However, some netizens revealed that teachers have already begun discussing whether using AI can help students learn.

GitHub offers 12 months of free Copilot access to maintainers of popular open source projects .

The criteria for popular projects are not clearly stated, but as long as you meet the requirements, you will see the free option when purchasing.

After 12 months, if you still meet the criteria, you can renew for another 12 months.

If you don’t meet the above conditions, Copilot is not the only AI code assistant on the market.

Tabnine also provides similar capabilities .

Compared to Copilot, Tabnine promises that user code will never be used for AI training and takes open source licenses into account in training data.

It also supports the use of your own code locally, training specialized AI models for specific programming languages, and supports offline operation.

Tabnine's basic version is free and can complete short code completion. The full-featured professional version costs $12/month.

Finally, after reading so many comparisons of Copilot’s usage experiences and pros and cons, do you think the price of 67 yuan/month for the official version is worth it?