aivs.tv

AI Coding

Voxel world, cloud model vs two DGX Sparks — Claude Sonnet 5.5 vs DeepSeek V4.1 Flash 1017

0:00 / 0:27
Posted byNiklas@niklaslenz_ai · on X
View post on X

7 likes1,397 views

Scoreboard2

  1. Claude Sonnet 5.5AnthropicClaude Code
    Time
    31.9 min
    Cost
    $4.60
    Tokens
    194.4K
    • Skyhaven; about $4.60 at API list prices, run on a subscription; browser check 9.0/10
  2. DeepSeek V4.1 Flash 1017DeepSeek
    Time
    1 h 28 min
    Cost
    —
    Tokens
    159.5K
    • Environment Local, two DGX Sparks
    • medium confidence
    • AURELIA; EXL3 2.9 bpw quant on TensorFold; browser check 8.5/10
    • Written “DeepSeek V4.1 Flash” after the 1017 update shipped; read as V4.1 Flash 1017.

Prompt

Voxel world, cloud model vs two DGX SparksThe post doesn’t quote the prompt wording.

Author’s verdict
The poster says the local world comes very close to the cloud one.

@niklaslenz_ai’s view (paraphrased)

Claude Sonnet 5.5 in the cloud and DeepSeek V4.1 Flash 1017 on two DGX Sparks get the same voxel-world prompt, one try each.

The post, itemised

Kind
Side-by-side
Task
Voxel world, cloud model vs two DGX Sparks
Category
3D & Shaders
Mode
One shot
Posted
Video length
27.3 s
Likes
7
Views
1,397
Bookmarks
1
Sponsored
No

NotesThe setups differ: the Sparks ran in a locked test container, Sonnet in the poster’s own CLI with internet access.

Cost & time, as stated in the postAs stated: Claude Sonnet 5.5 31.9 min, 194,375 tokens, about $4.60 at list prices; DeepSeek on two Sparks 88.1 min, 159,465 tokens, no API bill.

Related

Claude Sonnet 5.5DeepSeek V4.1 Flash 1017Claude Sonnet 5.5 vs DeepSeek V4.1 Flash 1017

Posted by @niklaslenz_ai on X. aivs.tv claims no ownership: it hosts a copy for fast playback and links to the post on X. Request removal

More with these models