In April 2024, the digital world witnessed the development of the world’s first AI software engineer, Cognition Labs’ Devin. Since then the race to build artificial intelligence-powered engineers began.
And now, a new and stronger contender has appeared in this race.
Recently, Cosine, a human reasoning lab co-founded by Alistair Pullen, unveiled Genie, an AI software engineer, which is regarded as the best in the world, surpassing even Devin.
Genie can autonomously solve bugs, build features, and refactor code. It scored 30.08% on SWE-Bench evaluations.
“We believe that if you want a model to behave like a software engineer, it has to be shown how a human software engineer works,” Alistair Pullen, Cosine’s CEO said.
This article will cover everything you need to know about Cosine Genie, its performance benchmarks, functionalities, and how you can gain access to it.
Scott Wu Net Worth: Devin AI Software Engineer, CEO of Cognition Labs
What is Cosine Genie?
Cosine Genie is an advanced AI software engineering model. It is designed to handle a wide array of tasks. Whether it is debugging, feature development, or code refactoring, Genie can do it all either fully autonomously or in collaboration with a user.
Cosine says “Genie is the world’s first AI Software Engineering colleague trained on data that perfectly emulates the cognitive processes, logic, and workflow of human engineers.”
Unlike other AI tools that only serve as copilots, Genie actively participates in the software development process.
Devin AI Software Engineer: How To Use It For Creating Websites, Apps, & Code Debugging?
Cosine Genie: Benchmark Performance
According to Cosine, Genie achieved the highest score among similar models on the SWE-Bench. The SWE is a comprehensive benchmark designed to evaluate the coding abilities of large language models across different software engineering tasks. Genie scored 30.08% and outperformed its competitors.
To put this in perspective, Devin, completed 500 out of 2,294 tasks, with a success rate of a mere 3.44%. Genie’s performance isn’t just a step forward; it represents a dramatic leap in the capabilities of AI within software development.
Core Functions
These are some of the key functionalities of AI engineer Genie:
- Autonomous Bug Solving: Genie can identify and fix bugs in your code without human intervention. It is expected to reduce the time and effort required for debugging.
- Feature Development: Genie can assist in the entire process, from planning to implementation. It can also help with adding new functionalities to your software or enhancing existing ones.
- Code Refactoring: The AI software engineer can clean up and optimize your codebase, making it more efficient and easier to maintain.
- Testing and Validation: Genie excels in validation by running comprehensive tests and analyzing outcomes to ensure the reliability of its solutions. It continuously iterates and improves until the desired results are achieved.
How to access Genie?
Currently, Genie is not available for public use. If you wish to access the AI software engineer, then you need to join this waitlist.
In the meantime, you can verify Genie’s success rate on GitHub. Cosine has made the final outputs of Genie publicly available on GitHub for independent verification. For more information, you can read the official technical report.
All You Need To Know About Devika, An Open Source Alternative To Devin
Future Prospects
Cosine recently acquired $2.5 million seed funding round. The funding round was led by U.S.-based venture capital firms Uphonest and SOMA Capital and also saw participation from investors like Lakestar, Focal, and others. Cosine is planning to expand its model portfolio to “include smaller models for simpler tasks and larger models for complex challenges.”
It would allow them to transform any state-of-the-art foundational model into a Genie model. The company’s plans include extending the context of an open-source model and pre-training a foundational model on their extensive dataset, to achieve enhanced generalization and better reconciliation of specialized data.
Alistair Pullen Net Worth: Co-founder & CEO of Cosine – AI Software Engineer
