Replit Makes AI More Accessible to Developers
Developer tooling vendor Replit is launching new initiatives to democratize AI development for all developers. With over 20 million users, Replit’s cloud software development platform has gained popularity in the developer community. Over the past year, Replit has been gradually introducing generative AI capabilities to its users through tools like GhostWriter AI code completion and partnerships with companies like Google.
Until now, access to GhostWriter has been limited to a subset of Replit users. However, this is about to change. Starting October 9th, Replit will integrate GhostWriter directly into its core platform, making the generative AI code completion tool available to all users. This move from Replit is being hailed as “AI for all.”
New Initiatives: GhostWriter Integration and replit-code-v1.5-3b
Alongside the GhostWriter integration, Replit also introduced a new version of its open source generative AI large language model (LLM) for coding – replit-code-v1.5-3b. This coding LLM is poised as a competitive alternative to similar models like StarCoder LLM and Llama CodeLlama 7B. Amjad Masad, CEO of Replit, emphasized the company’s mission to provide access to technology and empower developers worldwide, stating, “Our mission is to empower the next billion developers, and so we really didn’t want to create this world where some people have access to GhostWriter and other people don’t have access to it.”
AI in Developer Tooling
In recent years, AI-powered coding tools have become increasingly popular among developers. Major players like Microsoft, Amazon, and Stack Overflow have all introduced their own AI-assisted coding services. However, Replit has a different approach. Instead of treating AI as an add-on, Replit aims to integrate AI seamlessly into the programming experience. Masad believes that AI should be a default feature in Replit and other future products, making every programming interaction AI-infused.
With the new GhostWriter integration, Replit is transforming AI into a core feature available to all users, eliminating the need for a separate tool. Masad expressed his excitement about this deployment, stating, “We think this is going to be the biggest deployment of AI-enhanced coding in the world. We’re going to be burning as much GPU as we’re burning CPU, so pray for us.”
Replit’s Generative AI Capabilities
Replit’s generative AI capabilities are not reliant on another vendor’s LLM. Instead, the company has developed its own bespoke large language model. Michele Catasta, VP of AI at Replit, explained that the model was trained on open source code from GitHub, as well as code contributed by Replit’s user base. The release of the replit-code-v1.5-3b update signifies a significant expansion in the power and scope of Replit’s LLM. It was trained on 1 trillion tokens of code and supports 30 different programming languages.
Catasta highlighted the importance of data quality in training the model, as well as the powerful hardware used. Replit trained the model on 128 Nvidia H100-80G GPUs, which are highly sought after. Catasta stated, “To our knowledge, this is the first model officially announced to be trained on the H100 that has been released as open source, so we’re very excited about it.”
