DeepSeek V4.1-Flash-1017 inference benchmark in Sparkline: 64K prompt, 512 output

An inference benchmark of DeepSeek V4.1-Flash-1017 in Sparkline with 65,536-token prompts, 512-token outputs, concurrency 1 and 20 requests.

Sufyan

@sfxnzArchived 10 days agoOriginally posted 2026.10.21

DeepSeek V4.1-Flash-1017 inference benchmark in Sparkline: 64K prompt, 512 output

Sufyan @sfxnzOriginally posted 2026.10.21

Open original on X

Here is an inference benchmark running in Sparkline for DeepSeek V4.1-Flash-1017. Prompt length: 65536 tokens Output length: 512 tokens Concurrency: 1 Requests: 20

x.com/sfxnz/status/2112902787570786590Archived 2026.10.22 12:01

Open original on X