“If I process 500 million tokens on Claude, about $5,000. On our own model it's 97% cheaper.”
♫
Said on The Pomp Podcast
“The biggest mistake that people make is allowing it to get too deep. What you should be doing is compacting actively at 200, 300-ish thousand tokens.”
Dave Slutzkin · The Startup Podcast · Applied Wizardry: Cutting AI coding token costs, with Dave Slutzkin
Play from 12:03
12:03
Audio streams from the publisher’s own feed. Verbatim does not host or copy it.
What was described
Limit session depth to roughly 200-300k tokens instead of the million-token window
200,000-300,000
Said nearby on the same subject
“I have my auto-compact window in Anthropic models set to about 300,000 or 350,000 tokens, but I will compact manually over about 200,000.”
DS
Dave Slutzkin
The Startup Podcast ·
“I'm running a local model and I have 128 K tokens in my context window and that's working really well.”
♫
Said on .NET Rocks!
“AI changed that because it can distinguish between… words… and still get you a conclusion… instead of just looking for literally the word limited”
NC
Nick Coleman
Lawyerist Podcast ·
Moments like this, every Monday
The strongest new moments on your topics, in the speakers' own words. Free for three topics.
No card. No password. One-click unsubscribe.
Track a competitor?
Pro watches up to 25 names or phrases and emails you the morning after they are said.
See Pro, $29 a monthVerbatim quotes a short extract and links to the episode; the episode belongs to its publisher. Remove a show.