... Last time was voice, where one model has two input prices. Today goes back to caching, the pricing side of it. So what does caching

... mirande_ney): @farzam_habibi  the same context paid again on every turn, that's exactly the leak we see too... and yes the boring fix is always the one that works hehe. What did you use for the caching, the provider's prompt caching

... pages that never shows, because every request runs every middleware and overwrites everything. It matters on the group I built to be cacheable. My public pages run no session and no CSRF on purpose, so a shared cache is allowed to store them, and that means they skip exactly the middleware that would have overwritten the leftovers. A template there that printed the CSRF token would have printed the token of the previous visitor, into a cache

... audited what our own site actually answers, and the result came out backwards from what we assumed. A couple of hundred pages, most of them a template matrix built for search traffic. Those were cached properly. The pages the whole matrix funnels into, the homepage and pricing and the free generators and the legal pages, were answered no-store, every one of them. The pages we care most about were the only ones a shared cache

... finally read the whole study instead of the headline, and the actually useful findings are different from what got shared. First, in their realistic-config lane (instruction file + several MCP servers), the "light" harness came out HEAVIER: ~90.8k vs ~75k. A 72KB instruction file alone added ~20k tokens to every request, on both harnesses. Their own conclusion: configuration, not the harness, accounts for most of the production bill. The harness sets the floor, you set the ceiling. Second ... ... this is the one that changed how I think about it: the cache

... making over 35 redundant database queries on every single page load just to pull simple text strings. On a busy live server, this would cause a noticeable lag. To fix this, I decided to write a transient cache filter in my child theme's functions.php file. Instead of hitting the database every time a visitor opens the homepage, the theme now saves the query results in a temporary cache

... purpose. Each part of the app was built and shipped separately, and each one just reached for the same API the way the last one had, with no shared layer between them. Fixed it by keying a shared cache on the rounded coordinate pair and making every part check it before calling out, with a crash safe fallback if the cached

... change, the check didn't change, only the world underneath the cached input changed, and nothing re-read it. Do you think the fix is expiry, every escalation's premise gets a TTL and must be re-derived from source, not cache

... weeks ago, one of our developers pushed a refactor touching one of the agents Latitude runs in production. It looked harmless in review. Every request kept going through and latency stayed flat, but the prompt cache hit rate (how often an AI successfully reuses its cache