The speaker prefers Google Gemini LLM for coding due to its impressive token capacity and user-friendly features, finding it more effective than OpenAI’s ChatGPT and Anthropic’s Claude, which they found to be less reliable and intrusive. Despite some drawbacks, they advocate for Gemini as a powerful and cost-effective coding tool, encouraging others to consider it for their projects.
In the video, the speaker discusses their preference for using Google Gemini LLM for coding over other popular options like OpenAI’s ChatGPT and Anthropic’s Claude. They share their initial experiences with ChatGPT, which involved pasting code and errors to receive responses. While this method was effective, they found that ChatGPT could often provide incorrect answers, prompting them to explore alternatives. Claude was another option they tried, but its limited free plan and expensive upgrade options made it less appealing for their coding needs.
The speaker highlights the limitations of GitHub Copilot, which they found intrusive and complicated. They express frustration with its constant suggestions and the difficulty in getting it to understand their specific coding requirements. The speaker prefers a simpler coding environment, akin to Notepad++, and feels that Copilot’s integration into their workflow was more of a hindrance than a help. They also mention a previous project where they created a deep-seek powered IDE, which they found more user-friendly than Copilot.
The main reason the speaker favors Gemini is its impressive token capacity, allowing for a context window of up to 1 million tokens, compared to the 128,000 tokens offered by other models. This larger capacity is particularly beneficial for handling extensive codebases. They also appreciate the features of Google AI Studio, which allows users to add multiple files and provides a generous 2 million tokens in its experimental version. This flexibility enables the speaker to paste large amounts of code and receive relevant assistance without the constraints faced with other LLMs.
Despite their enthusiasm for Gemini, the speaker notes some drawbacks, such as its occasional inability to integrate with Google products effectively and its lack of proprietary knowledge. They recount experiences where Gemini struggled to provide accurate responses for specific coding tasks, leading them to revert to Claude for better results. The speaker emphasizes the importance of having a tool that can handle their coding needs efficiently without excessive costs or limitations.
In conclusion, the speaker advocates for Gemini as a powerful and cost-effective solution for coding assistance, especially for those who prefer not to have intrusive suggestions while coding. They encourage viewers who share similar frustrations with existing tools to consider Gemini for their coding projects. Additionally, they mention their plans to create a community for AI enthusiasts and entrepreneurs, inviting interested individuals to sign up for updates on upcoming courses and resources.