Leaderboard

DeepSeek V4.1 Flash (September)

Two DGX Sparksretired

On two DGX Sparks it decodes 81.9 tok/s. It gets 6 of 7 one-shot builds working. Ray tested it on 2026-10-17.

Machine
Two DGX Sparks: 2 × NVIDIA GB10, 2 × 128 GB unified memory, linked at 200 Gb/s
Engine
TensorFold Zig (Bertholomus AI)
Weights
EXL3 2.9 bpw community quant (Mia-AiLab)
Recipe
Tensor parallel over both Sparks, 1M context. Retired 2026-10-19 for 1017.
Made by
DeepSeek
Size
MoE
Tested
2026-10-17
Overall
—
Builds
857
Knowledge
—
Speed
618

Scores out of 1,000. Overall is the mean of the other three; how each is scored.

Speed

Decode speed
81.9 tok/s
Decode, prose
71.9 tok/s
Decode, Snake
91.9 tok/s
First token
339 ms
Prefill, 48k prompt
—
Tokens for the seven builds
225,888
Time for the seven builds
45m 50s

Decode speed is the median of a prose prompt and a Snake game, one request at a time. Prefill is how fast it reads a 48,000-token codebase before it answers.

Accuracy

SetCorrectScoreMinutes
MMLU-Pro, 140 questions, thinking low———
GSM8K, 100 problems, thinking low———
Snake by DeepSeek V4.1 Flash (September)
SnakeWorks
91.9 tok/s1m 22s7,546 tokens
Rocket launch by DeepSeek V4.1 Flash (September)
Rocket launchWorks
80.4 tok/s9m 57s47,975 tokens
Heptagon balls by DeepSeek V4.1 Flash (September)
Heptagon ballsWorks
85.4 tok/s3m 21s17,167 tokens
Solar system by DeepSeek V4.1 Flash (September)
Solar systemWorks
84.3 tok/s5m 05s25,694 tokens
Pelican on a bicycle by DeepSeek V4.1 Flash (September)
Pelican on a bicycleWorks
83.1 tok/s7m 08s35,514 tokens
City fireworks by DeepSeek V4.1 Flash (September)
City fireworksWorks
82.0 tok/s5m 41s27,992 tokens
No page written
Gummy bearsNo page
80.7 tok/s13m 13s64,000 tokens
No page: it was still writing at this run’s 64,000-token cap. The build was retired before the 262,144-token rerun that 1017 got.

Run local models in Ownlight

Ownlight connects the model servers you already run, like Ollama, LM Studio, Unsloth Studio, llama-swap or Model Deck, and starts Claude Code or Codex on a local model beside your cloud agents. Local is part of Free.

Get OwnlightDownload for MacHow Local works

Every step

Decode speed, time to first token, prefill speed, output tokens and total time for each prompt in the suite.

TestDecode tok/sFirst tokenPrefill tok/sOutput tokensTotal timeRun
Prose (speed)71.9339 ms —2,20931.1 s775
Snake91.9603 ms —7,5461m 22s741
Rocket launch80.4838 ms —47,9759m 57s795
Heptagon balls85.4424 ms —17,1673m 21s793
Solar system84.3402 ms —25,6945m 05s796
Pelican on a bicycle83.1352 ms —35,5147m 08s794
City fireworks82.0409 ms —27,9925m 41s792
Gummy bears80.7563 ms —64,00013m 13s784

More models on the bench

Updated 2026-10-20 16:39.