WHAT NEW IN X10.31 SAT

AI collections from X · 22 collections · 61,986 posts

deepseek-v4.1POST

DeepSeek V4.1 Flash gains traction on medical benchmarks at low cost

A user notes DeepSeek V4.1 Flash is trending for high medical benchmark scores at very low cost with MIT-licensed open weights, says the 1017 checkpoint surprised them in testing, but still picks Sol for quality-critical work.

Daniel Zambrini
@DanielZambriniArchived 4 days agoOriginally posted 2026.10.26
Daniel Zambrini @DanielZambriniOriginally posted 2026.10.26
Open original on X

DeepSeek V4.1 Flash is trending for its medical benchmark scores at a crazy low cost!! And it's open weights, MIT license, you can run it locally Cheap local models keep closing the gap. I still go to Sol when quality matters, but for the cheap tasks this is getting really hard to ignore The 1017 checkpoint already surprised me on my test, check it out:

This post quotes another post: “Retest after checkpoint mix-up: 1017 beats older V4.1 Flash on choreography and VFX”

x.com/danielzambrini/status/2114447070102794578Archived 2026.10.26 12:02