Deploy dedicated DeepSeek 32B on L40 GPUs ($8/hour)

Posted by wfalcon on Saturday February 01, 2025 @03:01PM
from the lightning.ai dept.

( 6 comments | 11 points | on HN )

Re:Deploy dedicated DeepSeek 32B on L40 GPUs ($8/hour) by yewnork on Saturday February 01, 2025 @03:58PM (#42902232)

curious the performance / price tradeoffs between deepseek-r1 671b, 70b, 32b

[ Parent | on HN ]

Re:Deploy dedicated DeepSeek 32B on L40 GPUs ($8/hour) by tchaton84 on Saturday February 01, 2025 @03:35PM (#42901988)

Does it support largest Deepseek model ?

[ Parent | on HN ]

Re:Deploy dedicated DeepSeek 32B on L40 GPUs ($8/hour) by neilbhatt on Saturday February 01, 2025 @03:16PM (#42901808)

nice, i can actually use my AWS start up creds

[ Parent | on HN ]

Re:Deploy dedicated DeepSeek 32B on L40 GPUs ($8/hour) by lmilad on Saturday February 01, 2025 @06:13PM (#42903522)

How well does DeepSeek R1 handle generating long pieces of text with Qwen 32B?

[ Parent | on HN ]

Re:Deploy dedicated DeepSeek 32B on L40 GPUs ($8/hour) by ashenWon on Saturday February 01, 2025 @08:47PM (#42904786)

Is this running ollama, vllm or sglang under the hood? Curious about these performance numbers.

[ Parent | on HN ]

Re:Deploy dedicated DeepSeek 32B on L40 GPUs ($8/hour) by woodr77 on Monday February 03, 2025 @04:06PM (#42923132)

Everyone's saying I needed H100s for this. L40 is way easier for me to get my hands on. great news.

[ Parent | on HN ]