[Aeolun]

I think the thing is that most of the people implementing stuff for Claude have already realized it’s just the best option available for… basically everything. I’ve switched to different models before, but I always come back to Sonnet or Opus for doing anything sensible.

[dgunay]

Claude may be arguably the best model, but why decide unilaterally for your users that they _have_ to use it?

If there's no particular feature that only Claude offers, this is just needless vendor lock-in. And what happens if another lab releases a model that suddenly trounces Claude at coding? Your users will leave for an app that supports the new hotness, and you won't be able to keep them because of a short-sighted architecture that cannot swap model providers.

[int_19h]

The situation with models right now is that to eke out that last bit of performance, you have to do some things in ways that are specific to the model in question - wording of prompts, when and where to introduce relevant parts into context etc.

[rsanheim]

Because Claude code does offer particular features. More important than “features” is the fact that it works and does the things you want like 60-70% of the time, with guardrails and practice and attention. Which is way better than competing tools.

Besides that. These tools are changing so fast that to build an agent agnostic tool would be insane given the speed and market pressures right now. Why support roo or cline or cursor cli if it adds 3x engineering cost for 20% more market reach? The reality is there are no standards around the way the actual leading tools work if you wanna build something on Claude/codex/(insert flavor of the week).

Gotta pick your horse and try to hang on, and hope you picked right.

[Aeolun]

> Claude may be arguably the best model, but why decide unilaterally for your users that they _have_ to use it?

What a ridiculous proposition. It’s me making the app right? You getting to use it (if you want) is purely incidental. If you never use it because it doesn’t support anything but claude, that’s not something I consider a problem.

[threecheese]

There’s value in Anthropic being able to optimize their model’s front end to fulfill whatever features they plan for CC - like tool calling. You point out a valid risk of lock-in, maybe this signals they are committed to being at the forefront of coding models (part of enterprise play)?

[rblatz]

While I prefer Cline/Roo at work where I have multiple API plans for AI models, for personal I have Claude Pro and that really only works with Claude code. The benefit is that I can use it on a $20 a month plan.

[commandar]

Roo can use Claude Code as a provider.

I mostly use Claude Code with a Max plan via Roo. I have the option of sending prompts to OpenRouter if I've hit usage limits or if I want to try a particular task with a different model (e.g., I'll sometimes flip to Gemini Pro if a particular task could benefit its large context windows).

[dottjt]

I wonder if LLMs are actually closer to programming languages, in the sense of how they'll proliferate amongst different companies/people/use cases. Like maybe OpenAI is considered the Java of LLMs, while Claude is more like Python etc.

[block_dagger]

ChatGPT feels more like CoffeeScript to me - theoretically a good idea at the time but bound to be replaced.

[kristo]

This is true for now.

I built some workflows using Claude’s API and now wish I had used a wrapper so I could easily switch to try gpt-5 for the cost savings.

[cameronh90]

I'm finding GPT 5 (via Codex CLI on a pro subscription) is far better than Opus for my use cases. Much more than the small difference on swe-bench would suggest. However, the Codex CLI is so immature by comparison that I'm still mostly using Claude, and only escalating to Codex when Claude snookers itself.

[kordlessagain]

I've been using Codex in another tab in Terminal on Windows and it's my go to agent now. Just my two cents. I have a lot of hours with Claude Code, and do appreciate it, but Codex is quite good.

[ubercow13]

Claude doesn't seem very good at translation to me compared to GPT*. It also can't understand video like Gemini.