Post #1986380

2026-04-23 11:23 UTC

@ChadF @dave Yeah, I've always had it on the highest. I tried to put it on medium but it still burnt extremely fast. I even had to add $40 of extra credits to finish a work thing, and it basically burnt through them in 20 minutes 😂 Every time I refreshed the usage page I could see it had used $2. If this is basically the "real cost", then I think the whole AI card house might come crashing down at some point, and people look towards local or non-frontier models for a lot of work.

Replies (4)

@ChadF@podcastindex.social 2026-04-23 11:30

@martin @dave I noticed I had $100 in overage credits so I've been trying to hit it hard to burn through those before it's too late.

Open ##1986381
@chrislw@podcastindex.social 2026-04-23 11:41

@martin @ChadF @dave when the companies IPO, shareholders will not accept the lack of profitability that investors do. When the costs go up across the board, and as more cases of hallucinations in fields like law keep coming up, things will not go well.

Open ##1986386
@nathan@xoxo.zone 2026-04-23 14:05

@martin @ChadF @dave Theo Browne did a deep dive into the overlapping factors causing CC to perform worse. - Tokenizer Changes in Opus 4.7 - Default 1 Million Token Context Window - Thinking Content Redaction - Increased API Thrashing - Default Lower Effort Level The only 2 things in your control are the default effort level and adding `CLAUDE_CODE_DISABLE_1M_CONTEXT=1` to your `.claude/settings.json` https://www.youtube.com/watch?v=KFisvc-AMII

Open ##1986387
@ville@podcastindex.social 2026-04-23 17:02

@martin @ChadF @dave what if this isn't yet even near the real price? 😬

Open ##1986395