DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression · HackerPapers discussions

DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression — HN submission

Wed, 28 Oct 2026 19:00:25 GMT

DeepSeek-v4.1-Flash: Pushing the Limits of KV Cache Compression

6 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression — mentioned by dormento

Mon, 26 Oct 2026 16:37:02 GMT

There's also that paper "DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression" which got released some time ago. My take is the breakthroughs got quickly implemented and helped drive the inference prices down. Edit: link to the paper https://arxiv.org/abs/2609.19969 (it was published 2026-09-17)

Paper and discussion history. A mention is not an endorsement.

DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression — mentioned by gpugreg

Fri, 09 Oct 2026 17:58:55 GMT

> Deepseek also came with heavily subsidized plans, at least at first. Do you have a source for that? I got a source straight from the hoses mouth, which claims that DeepSeek-R1 was highly profitable: > If all tokens were billed at DeepSeek-R1’s pricing (*), the total daily revenue would be $562,027, with a cost profit margin of 545%. https://github.com/deepseek-ai/open-infra-index/blob/main/20... With DeepSeek-V4.1-Flash, the memory requirements for the KV cache have been reduced by over 50x and FLOPS by over 5x, so it is even cheaper: https://arxiv.org/pdf/2609.19969

Paper and discussion history. A mention is not an endorsement.