Google recently made news with a new Gemini release. The coding performance is very decent, but where it really shines is the speed (tokens per second) and price. My workflow so far was to work with a combination of ChatGPT o1 for creating the prompt → Cline with Claude 3.5 Sonnet. That works well, but it's expensive. o1-pro is $200 per month and Anthropic charges you $15 for 1M tokens. Yikes. Compared to that at $0.40 per 1M tokens is a fantastic deal. It also has an extremely large 1M tokens context window, meaning you can fit your whole codebase into its "memory". If you want to check it out for free (with some restrictions), you can do it in the AI Studio. I'll be plugging it into Cline and testing it out. A couple days ago the experimental Gemini 2.0 Pro also got released to the public, which is their best coding model. Exciting times and it looks like Google's not to be beaten by the AI startups, which makes sense, considering they have all the data to train the models and the custom hardware to run it on. Yours, Taj |
Hi, I’m Taj Pelc. Building for the web and leading engineering teams for 15+ years.