Hands-on: DeepSeek V4.1 Flash is fast but overthinks, one task took over an hour

J A Z I I @notjaziiOriginally posted 2026.10.17

Open original on X

deepseek v4.1-1017 is super slow token speed is crazy good, but model spends so much time overthinking and testing that a single task takes hours it's a great model and mogs most chinese models and even opus 5, but team really needs to fix it this one test took over 1 hour and only cost $1. looks kinda good, but nothing impressive for that much waiting how's your experience with this model so far?

This post quotes another post: “Gemini 4 Argon vs DeepSeek v4.1: same-prompt render test”