alt Hacker Newsalso for economic reason. Anthropic design Claude code so they are in control regarding how token get burned. To a degree.
With 3rd party, designers have not incentive to be token consumption conscious.
Gemini has a 1M token context, flash can be used for free via the web interface, can't paste more than handful thousands token.
Replies
raincole • yesterday at 8:56 PM
> With 3rd party, designers have not incentive to be token consumption conscious.
How? If I used open code I'd hope it to not eat tokens unnecessarily so my subscription quota lasts longer.
I add 'only do this if...' to every single skill I write. It's a bit of voodoo-style wishful thinking, I admit.
Agreed. What I suspect is: the dynamic model routing on CC is way stronger than people realize, and that "Percent-based usage" is intentionally vague because while it is probably measuring "200M tokens per week" or something, they don't want you asking questions about whether you're getting 200M Haiku tokens or 200M Opus tokens. A token is a token to the usage limit, where it comes from doesn't matter to the usage limit. But, to OpenCode it might, because OpenCode can just fire-and-forget everything at Opus (and probably does).