Post

Madza 👨‍💻⚡

Which LLM model has been the best for coding for you? A: Claude Sonnet 4.5 B: Claude Opus 4.5 C: Gemini 3 Pro D: GPT-5.1 Codex

26/11/25, 1:19 am0 reposts1 like

@alannewman.bsky.social

Recently Codex.

Madza 👨‍💻⚡

Thanks for sharing!

@alannewman.bsky.social

Yeah, I’ve paired it with V0 for an MVP I am building. For Codex what is working really well is discussing what I want to build with Chat GPT and then asking it to write the prompt for Codex. Codex more often than not one shots each feature I prompt this way.

Jules - revenue-driven SEO

@julesignaciocanlas.bsky.social

Claude Sonnet

Madza 👨‍💻⚡

Thanks a lot for sharing, mate!

Jules - revenue-driven SEO

@julesignaciocanlas.bsky.social

You're welcome😊 What's your pick?

Madza 👨‍💻⚡

Latelly been testing Opus 4.5 a lot and getting great results for coding.

Boon aka Hwee-Boon Yar

Opus 4.1. Undecided about 4.5 still.

Madza 👨‍💻⚡

same here, also 4.5 is amazing

Boon aka Hwee-Boon Yar

I think if they had maintained the Max limits while it was 4.1, I'd been happy with Opus 4.1 for the longest time.