Comparing AI Code Generation Tools
Since the arrival of ChatGPT, there is a tremendous increase in AI tools which can help you in your daily chores starting from answering a simple question, audio/video generation, legal documentation, chatbots to code generation.
Yes, you heard it right you can generate good amount of code using these code generation tools and save fair amount of time.
Few days back I got to compare the AI code generation tools. So I compared 3 tools out of many.
I will not take you in the details like since when these tools are available and on which model they are built etc. I will only highlight my observations with these 3 tools in this blog and then you can decide which one is the best suited for your need.
Comparison:
- Free Tier:
- Github Co-pilot does not provide any free tier. 😐 You have to buy subscription plan only to check the suitability to your projects.
- TabNine provides a free tier where you can use very limited set of features and evaluate the tool.
- AWS CodeWhisperer provides free tier for individual developer and provides full capabilities in the free tier also.😍
2. AI Capabilities
- Github Co-pilot is designed more for general purpose development. The main advantage of co-pilot is that it’s trained on a very large set of code base by Microsoft.
- Tabnine is also designed for the general purpose development but it learns from your code base and usage patterns.
- AWS CodeWhisperer — AWS does not disclose on which model it’s trained but it has best support for AWS related services and code generation. I have also experienced that AWS CodeWhisperer very good on general purpose code development too.
3. Supported IDEs
- Github Co-pilot supports JetBrains and VisualStudio
- Tabnine supports wide range of IDEs you can see here
- AWS CodeWhisperer supports- AWS Sagemaker Studio, JupyterLab, VS Code, JetBrains, AWS Cloud9. You can see more details here.
Similarities:
- They all offer wide use case of code generation and they are queally good. With all these tools you basically define what you want to do in the comment and it will suggest
- Co-pilot and CodeWhisperer are both cloud based solutions
- None of these tools claim that they can generate the entire software. They just aid you in generating the code snippet which you can re-use.
My choice:
I ended up with selecting AWS CodeWhisperer as I write many serverless code in AWS Lambda and also do development in typescript and front-end.
AWS CodeWhiperer gives you awesome capabilities to generate Lambda and AWS SDK methods. For example you need to generate the method to pull the files from S3, just method put some method name relevant like getFilesFromS3Bucket and it will create entire method for you.😃
Further, when you go to your lamdaHandler, it also provides you the suggestions to add the method calls which you have generated using the AWS CodeWhisperer. 😎
Closing Words:
The right selection of the AI tool for you totally depends on your use case and also, the license cost you are ready to bear 😉.
