Google has made a significant move in the world of AI coding tools by releasing a free version of Gemini Code Assist. Previously available only on paid business plans starting at $19 per month, this powerful AI coding tool is now accessible to individuals without any cost. With this release, users are allowed up to 180,000 code completions per month, a substantial increase compared to the 2,000 code completions offered by its main competitor, GitHub Copilot.
In a recent discussion with Ryan J. Salva, a senior director of product management at Google Cloud, insights were shared about this exciting development. Before joining Google, Salva held a pivotal role at GitHub, which places him in a unique position to compare Gemini Code Assist with GitHub's market-leading tool.
Salva explained that the decision to offer 180,000 code completions was to meet the needs of 98-99% of developers. "We wanted to effectively offer a service that met the needs of 98-99% of developers who were banging on it all day," he stated. This generous cap is designed to be practically unlimited, accommodating even the most dedicated engineers.
Despite the increased burden on Google's AI compute infrastructure, Salva emphasized Google's expertise in handling massive scale, drawing parallels to the successful scaling of Gmail and Google Search.
The free version of Gemini Code Assist is powered by the latest Gemini 2.0, positioning it on par with GitHub Copilot, which uses OpenAI's GPT-4o. Notably, Gemini Code Assist supports all programming languages in the public domain and offers a 128,000 token limit for chat in its free tier, compared to GitHub Copilot's 64,000 token window.
Another key feature is the new code review agent available as a GitHub app, called Gemini Code Assist for GitHub. This highlights Google's commitment to offering comprehensive AI coding solutions.
While the free version is impressive, Salva pointed out additional capabilities available on paid plans, such as the ability to pull data from remote repositories. Unlike GitHub, which encourages users to stay within its ecosystem, Google aims to connect with a variety of source control management solutions, offering flexibility to organizations using platforms like GitLab and Bitbucket.
The release of Gemini Code Assist comes at a time when numerous new AI coding tools are entering the market. These tools, often dubbed agentic IDEs, have the potential to create entire applications from scratch. Notable examples include Bolt, Windsurf, and Lovable, which target both professional and amateur developers.
Salva indicated that Google is keen on tapping into this market, emphasizing the importance of natural language as a foundation for interaction. "Our general approach is to think about the requirement doc and natural language as being the foundation," he explained, highlighting a shift towards a more intuitive way of developing software.
Looking ahead, Salva assured that Google is just getting started with AI-assisted coding tools. The current release is seen as a foundation for future developments, which will include new agentic capabilities and the integration of AI beyond traditional IDEs.
As Google continues to innovate in this space, developers, both professional and amateur, are encouraged to explore the new Gemini Code Assist and leverage its extensive code completion capabilities.
