Hacker News: hirak10https://news.ycombinator.com/user?id=hirak10Hacker News RSShttps://hnrss.org/hnrss v2.1.1Mon, 31 Aug 2026 08:15:45 +0000<![CDATA[New comment by hirak10 in "Launch HN: Speko (YC S26) – OpenRouter for Voice AI"]]>The piece I'd want to see in the constraint solver is effective cost rather than list price. On the LLM leg alone, current 5.6 pricing splits at 272K input tokens with the long band at exactly 2x, and cached input runs 90% below standard input, so two stacks with identical list prices can differ several-fold depending on how much of the prompt is a stable prefix.

Does the optimizer model cache-hit rate and context distribution, or does it score on list rates?

]]>Tue, 18 Aug 2026 22:24:47 +0000https://news.ycombinator.com/item?id=49353640hirak10https://news.ycombinator.com/item?id=49353640https://news.ycombinator.com/item?id=49353640