Hacker News Reader

Comment by satvikpendem | Hacker News Reader

No, it's more that Gemini models are simply not very good for coding compared to the top two. Even with Antigravity I use Claude models.

View on HN

Gemma 4 31b is better for coding than Gemini in my limited testing on a small C project single source file project, less than 1000 lines. Setting temperature to 0 gives better results for me. It seems like Gemini ignores the system prompt more and the default reasoning output seems more incoherent.

> Gemma 4 31b is better for coding than Gemini

Is there a fine-tuned Gemma coding model? I'd assume that would perform quite well.

I don't know, but for me setting temperature to 0 has a noticeable impact. I don't think you need a fine tune.

Their open weight on device models are really impressive. Partly because I think they are the only ones out of all the frontier labs even working on local models.

Depends on the language. Gemini and Claude are far superior when it comes to C# for instance, compared to anything that OAI offers.

Yaah I feel the same way. Gemini is great at and Django and AI backends, OpenAI better at making something visually pleasing in React and Claude for everything else or across frontend and backend.

At least, that's my heuristic that tends to work for my workflow. I use a combination of Gemini-CLI, Claude Code, and Github Copilot, but across those, the underlying model choice works best according to which part of the applicaiton I am messing with

Reply on news.ycombinator.com