I wrote an article before introducing AI coding " It's okay if you are not familiar with the code, let AI write it for you ". They are all good at generating code), and then introduce some independent vertical applications. They are more focused on the programming field, providing independent tools or IDE plug-ins, which actually help programmers improve the efficiency of producing code.
Github Copilot
https://github.com/features/copilot/
GitHub Copilot is trained on billions of lines of code to turn natural language prompts into coding suggestions across dozens of languages. Backed by the two big trees of Microsoft and OpenAI, as well as the unique massive code base of Github, the ability is extremely powerful, and of course the cost is quite high.
Codegeex2
https://codegeex.cn/zh-CN
CodeGeeX comes from the Tsinghua Zhipu team in China. It is a multi-programming language code generation pre-training model with 13 billion parameters. CodeGeeX is implemented using the Huawei MindSpore framework and trained on 192 nodes (a total of 1536 domestic Ascend 910 AI processors) in the Pengcheng Lab "Pengcheng Cloud Brain II". As of June 22, 2022, CodeGeeX has been pre-trained on code corpora (>850 billion Tokens) in more than 20 programming languages for two months.
CodeGeeX2 is implemented based on the ChatGLM2 architecture and added code pre-training. Thanks to the better performance of ChatGLM2, it is open source and free. I believe that friends who write codes are already using it. It is generally believed that the efficiency of writing codes has increased by more than 2 times.
Cursor
https://www.cursor.so/
As the first batch of OpenAI investment incubation companies, Cursor has unique advantages. Cursor is an independent IDE tool, and its future ambitions can be seen.
Small
https://bito.ai/
Recently, it received a financing of 3.2 million US dollars, and the API service behind it is also connected to OpenAI, but I believe that he has made another layer of integration on the outer layer, which has a relatively market among domestic programmers.
CodeWhisperer
https://aws.amazon.com/codewhisperer/
A complete latecomer, Amazon has not launched any major products in the field of AI large-scale models, which has a certain relationship with its positioning: an e-commerce + cloud service manufacturer, with a slightly delayed sense of technology strategy.
Released in preview as part of the AWS IDE toolkit, meaning developers can use it in their IDEs, including Visual Studio Code, IntelliJ IDEA, PyCharm, WebStorm, and Amazon's own AWS Cloud 9.
cost comparison
Expenses are an item that everyone is generally concerned about. After the efficiency is improved, it will be painful for individuals to pay a fee. It doesn't matter if the company pays.
| name | personal use | advanced version |
|---|---|---|
| CodeWhisperer | free | $19/month |
| Codegeex2 | free | free |
| Github Copilot | $10/month | $19/month |
| Cursor | limited use | $16/month |
| Small | free | custom made |
No matter which AI editing assistant, programmers have experienced from manual coding, to low-code, zero-code, and now to the era of AI-generated code. The times are advancing, and the tedious and non-technical Coding is finally completed by machines. Let's engage in more thoughtful work and be a coachman who actively embraces change.
The official account replies " Claude actual combat ", " ChatGPT actual combat ", " WPSAI actual combat ", and obtains the corresponding e-books.
—Extended reading—
WPS Office AI combat summary, the era of intelligent office has come
After interviewing dozens of middle and high-level companies, you must understand these things
AI large-scale model applications that are taking place or are about to take place