fmerian

What's the best AI model for coding?

by

New AI models pop up every week. Some developer tools like @Cursor, @Zed, and @Kilo Code let you choose between different models, while more opinionated products like @Amp and @Tonkotsu default to 1 model.

Curious what the community recommends for coding tasks? Any preferences?

3K views

Add a comment

Replies

Best
Teofilo Rassin

At this stage I pick based on trust + explainability. Being able to trust the suggestions and understand the reasoning is more valuable than whichever one has the highest benchmark score.

Mihir Kanzariya

Cursor or claude with blocpad

Zaid

I use Antigravity from google, and mainly use Sonnet and Opus. Its weird but on the same prompt and context, sometimes, Gemini models hallucinates and start rewriting the whole files

Rajat Jain

Claude Opus 4.5 (Thinking) 32K ?

Sacha MORARD

Anthropic has a huge head start with their models. But to be precise, it's the models coupled with Claude Code that allow us to enter a new era and give superpowers to developers.
The only downside to all this remains the price of tokens which are starting to become more and more uncontrollable.

I work a lot on this issue, particularly on token compression technology, which will allow us to better reduce, control, and make the ecosystem a bit more frugal... at least I hope so.

fmerian

@sachamorard - could @Edgee help us reduce the CC bill?

Sacha MORARD

@fmerian Absolutely! That's what we do. @Edgee is a Token Compression technology which can save up to 50% of the token bill. And it works with Claude Code pro/max plan as well

Dima Hanzhelo

Why have you missed the Opus 4.5?

fmerian

good point - can't edit the poll! +1 @Claude by Anthropic

David
From left field, and after a lot of experimentation, I'm currently going to swear by; cosine.sh hooked with cubic.dev to catch any bugs.
fmerian

@iamasuperuser oooooh good point. I forgot to include Genie 2 by @Cosine in the list. heard great things about @cubic (currently using @OpenAI and @Claude by Anthropic according to their docs). great combo!

JohannKrugell

Sometime Grok for a good outline and Sonnet 4.5 for the detail work using Cursor as the IDE

fmerian

@johannkrugell any experiences with Composer 1?

Jannik Jung
For me personally: gpt-5.2-codex all the way. Dry as a brick, but the results speak for themselves.
Elvis Baugh

I prefer Opus 4.5 more than Sonnet. Even though Opus maybe be a tad bit slower but it provides better reasoning and context understanding. I personally feel Opus provides a balance in that sense.

fmerian

@Claude by Anthropic family (Sonnet, Opus) is definitely leading the way