Commit

b3f2275

I've spent a lot of time refining my Claude Code workflows, and it's honestly made a huge difference. My productivity has skyrocketed, regressions are pretty much a thing of the past, code quality is consistently high, security is solid, and everything is well organized (shoutout to my pre-AI dev experience and deep backend knowledge). But the downside is how heavily I rely on Claude, and how much I consume. I'm getting increasingly frustrated with the x5 usage limits. I could easily do so much more with x10. But even then, I'd still be hitting limits eventually. Open-weight models are really interesting, but they're just not cost-effective for my use case. Hardware is ridiculously expensive, and cloud solutions like DeepSeek can quickly end up costing more than a Claude subscription. Does anyone else feel like they're reaching a point where they just can't tolerate being dependent on usage quotas anymore?

commit b3f2275

The prices will only keep going up, as well. Price increases and shrink-flation are already hitting LLM plans across the board. It will be pushed as far as they can take it, in an effort to reach profitability. How much % of an equivalent human wage are people willing to pay…
i genuinely feel a great breakthrough will come which will make these models really cheap, in the 1950s, storing a few megabytes required enormous, expensive equipment. Today, a 1 TB SSD is an ordinary consumer product