Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

5 Things to Try with Gemini 3 Pro in Gemini CLI

developers.googleblog.com

Post reply on HN
Print view View original

41–48 of 48 posts

Re: 5 Things to Try with Gemini 3 Pro in Gemini CLI

#41

Post by NamlchakKhandro »

esafak wrote: post #17
chis wrote: post #3

Has anyone switched to Gemini CLI? It's so important but also exhausting keeping up with which model is the leading edge. Especially since every model has different idiosyncrasies you have to learn to work with it effectively. Currently my ranking is * Cursor composer: impressively fast and able but not tuned to be that agentic, so it's better for one-shot code changes than long-running tasks. Fantastic UI. * Claude…

I think the TUI agents are pretty similar; I can use CC, Codex, Gemini, and opencode interchangeably.

once you have opencode, why would you bother with any of the others?

Re: 5 Things to Try with Gemini 3 Pro in Gemini CLI

#42

Post by NamlchakKhandro »

chis wrote: post #3

Has anyone switched to Gemini CLI? It's so important but also exhausting keeping up with which model is the leading edge. Especially since every model has different idiosyncrasies you have to learn to work with it effectively. Currently my ranking is * Cursor composer: impressively fast and able but not tuned to be that agentic, so it's better for one-shot code changes than long-running tasks. Fantastic UI. * Claude…

why would you bother with any of these when opencode exists?

Re: 5 Things to Try with Gemini 3 Pro in Gemini CLI

#43

Post by NamlchakKhandro »

nateb2022 wrote: post #37
RealityVoid wrote: post #30

Earlier quoted context omitted.

Excuse my ignorance, but... What? It runs cli commands, why would the language it's written in matter for the integration? What am I missing here?

From a practicality standpoint, writing in TS allows you to execute arbitrary JS in-process with a simple exec() call; I have no doubt they'll use this to more deeply integrate agents with existing codebases in the near future. E.g. agents rather than just reading the code, will be able to directly import the data structures themselves, use libraries within the JS/TS ecosystem to parse code into an AST, and execute i…

no. i think your reasoning is flawed.

think about how it would do the same for python/golang/lua/etc.

I would never write a TUI like this and simply `eval` in the same scope as the TUI frames.

you shell out and run it with the language engine required.

Opencode has the ability to lean on LSP and Formatters, but these are not required to be written in the same language as the TUI

Re: 5 Things to Try with Gemini 3 Pro in Gemini CLI

#44

Post by NaomiLehman »

NamlchakKhandro wrote: post #42
chis wrote: post #3

Has anyone switched to Gemini CLI? It's so important but also exhausting keeping up with which model is the leading edge. Especially since every model has different idiosyncrasies you have to learn to work with it effectively. Currently my ranking is * Cursor composer: impressively fast and able but not tuned to be that agentic, so it's better for one-shot code changes than long-running tasks. Fantastic UI. * Claude…

why would you bother with any of these when opencode exists?

I find Claude Code a bit better than opencode. I think it's not only about the model but orchestration/context handling.

Re: 5 Things to Try with Gemini 3 Pro in Gemini CLI

#46

Post by api »

NaomiLehman wrote: post #45
eptcyka wrote: post #32

Why would you `git bisect` when you can `git log -p` and search for “dark”? Why is a marketing listicle on top of HN?

half of hackernews is AI-generated blogs and the other is marketing material. we have to accept it at this point.

That’s better than most other social sites.

Re: 5 Things to Try with Gemini 3 Pro in Gemini CLI

#48

Post by pshirshov »

nateb2022 wrote: post #26
embedding-shape wrote: post #14

Earlier quoted context omitted.

I haven't tried Gemini CLI with Gemini 3 Pro, but pretty much all the others. I usually run four agents at the same time, for each task, giving them the same prompt and then comparing their responses. Gemini CLI has the lowest rate limits, lowest inability to steer the models (not sure that's a model or tooling thing, but I cannot get any of the Google models to stop outputting code comments constantly and everywhere…

> Gemini CLI has the lowest rate limits For Gemini 3.0, the rate limits are very very generous. Google says rate limits refresh every five hours, and that only “a very small fraction of power users” will ever hit the limits.

I hit them constantly on AI Ultra subscription.
Post reply on HN