DeepSeek V4–almost on the frontier, a fraction of the price

Re:DeepSeek V4–almost on the frontier, a fraction of the price by halfwhey on Saturday May 02, 2026 @09:09AM (#47986104)

Might be a dumb question but do you have to read the files in the same order in new sessions to ensure the correct prefix for the cache?

[ Parent | on HN ]

Re:DeepSeek V4–almost on the frontier, a fraction of the price by WatchDog on Saturday May 02, 2026 @09:38AM (#47986304)

Yes, you have to use the same session, I guess you could load up a bunch of context, then fork the session into a few different tasks, although I haven't tried it.

[ Parent | on HN ]

Re:DeepSeek V4–almost on the frontier, a fraction of the price by weiliddat on Saturday May 02, 2026 @09:34AM (#47986274)

Also curious. With tool calls reading/searching different files, possible compacting reading a large codebase / long threads, I can't imagine how you hit 99% cache rate.

[ Parent | on HN ]

Re:DeepSeek V4–almost on the frontier, a fraction of the price by naaqq on Saturday May 02, 2026 @09:35AM (#47986284)

Sorry, I was wrong here. I meant a single long session. And there’s no compression, the 1M context is only half used.

[ Parent | on HN ]

Re:DeepSeek V4–almost on the frontier, a fraction of the price by gbgarbeb on Sunday May 03, 2026 @12:29PM (#47998604)

Then where did 200M come from? 200,000 tokens?

[ Parent | on HN ]

Re:DeepSeek V4–almost on the frontier, a fraction of the price by naaqq on Monday May 04, 2026 @03:56AM (#48005877)

Not all read tokens are included in the context, many of the tokens are from read cache hits. I hit it many times so it grew to 200M. The number came from the API platform.

[ Parent | on HN ]