Did GitHub Just Kill Cursor? (NEW Copilot + Spark Update 2024)

Key Takeaways

did GitHub just kill cursor AI GitHub just announced two massive updates that are going to change the game for AI generative coding let's break them down here so GitHub just finished hosting their Universe conference which is basically just a global event that's focused for developers in the ecosystem and the event focused primarily on new AI driven developer tools now the first massive news announcement is that github's co-pilot now offers multi llm model support which they called developers Choice the new models will be first rolling out in co-pilot chat so that you can test it out for yourself and see if it's a good direct competitor to cursor AI now here's why this is massive co-pilot started exclusively with open AI so they first had codex and then they moved on to GPT 3.5 and then the model we all know and love GPT 4 this exclusive relationship was one reason why tools like cursor gain so much traction within the space they offer choice and freedom with other llm models particularly Claude which was Superior to coding in every way especially when compared to the other open AI models that were existing at the time but now GitHub just dropped this bomb to really stir up competition within the space they've added anthropics Claude 3.5 Sonet model Google Gemini's 1.5 pro model as well as open ai's new 01 mini and 01 preview models now GitHub offers the same exact flexibility that cursor provides plus it's already integrated into the GitHub ecosystem which is where most of us already work within and with this update each model brings something unique to the table CLA 3.5 Sonet is still perfect for complex coding tasks Gemini's 1.5 pro has a massive 2 million token context window and open ai's new 01 model provide better reasoning for edge cases and if that wasn't all GitHub co-pilot is also partnering with perplexity the market leader in AI search this means that you can ask perplexity any question pertaining to your code base or just to stay up to date with the framework that you're using now the second massive news update from this GitHub Universe event is the unveiling of GitHub spark GitHub spark is an AI native tool designed to bring app creation to everyone now complete beginners to Pros can create these Sparks which are basically micro applications these micro apps are created using only natural language prompts meaning that it truly is no code and as intuitive as it gets all you need to do is type in a prompt and these Sparks are immediately created you don't write or deploy any code at all in fact you don't even see any lines of code and the best thing is that with GitHub spark you'll be able to see instant previews as soon as you finish typing up your prompt and then to finalize the spark you'll have instant automatic deployment to github's environment so to summarize GitHub Sparks Works through three main components first is a natural lay language based editor where you can describe your ideas to initially create the spark as well as refine your spark over time the second is the manage runtime environment which hosts all the Sparks that you make providing access such as data storage theming and llms the third is a pwa enabled dashboard which lets you manage and launch the Sparks that you made from anywhere and these Sparks can be created from four different llm models so first we have cla's 3.5 Sonet model GPT 4 as well as the two latest models from open AI 01 mini and 01 preview now here's the key thing about GitHub spark these Sparks are built around the idea of micro apps which are focused apps that only do one singular thing very well GitHub stress that these Sparks are intended to be single-purpose tools meaning that you won't be able to build any complex applications through these Sparks they should only be doing one thing very well whether it be study Flash cards or an allowance tracker for example and these Sparks all run within github's managed environment meaning that you won't be able to deploy these Sparks anywhere else at least for now so you shouldn't think of these Sparks as being somewhat of a competitor to what cursor AI would help you with given that it's not intended for complex applications you can't dig deep into the code these are more so just simple prompts that are either for personal tasks or they can also be used for something like an autonomous agent so you can use these Sparks to create things such as personal tools internal tools team productivity apps as well as quick prototypes for ideas that you have and you'll still be able to share these Sparks as well so you can invite others to use your app directly you can also have others remix your app but ultimately you control the read and write permissions for your spark and everything stays within the GitHub ecosystem so I can see this ending up being one of the easiest ways to create an AI autonomous agent for any situation or need regardless of your technical level and now as a final note and possibly the most important is the fact that these Sparks are not yet released so there's currently a sign up list for the beta version of these GitHub Sparks but as of now there's no public access to use these GitHub spark features now here's my honest take on these latest updates copilot's multimodel approach is a direct challenge the cursor offering the same model flexibility the same core functionality as cursor composer with multifile line editing as well as native GitHub integration I'm personally watching and waiting for cursor's next move as I wouldn't count out that talented team just yet but I do think this is a very exciting development that promotes healthy competition within the space so what do you think do you think that co-pilot's multimodel approach will make you reconsider using cursor do you plan on using spark for any of your personal tools drop a comment below and let us know don't forget to like And subscribe for more AI news and if you're looking for a program that will help you build apps completely with AI with real-time developers to help out when AI gets stuck make sure to check out our active community at AI code pathways