Head to head

Any two models from the board, on one scale.

Model A

Model B

openrouter · dots-studiohosted

overall

20 ▸

openrouter · deepseekhosted

40vs60

DeepSeek V4.1 Flash leads by 20 overall · ahead on 9 of 9 tasks

tasks

Task by task

9 tasks

66TOOL+12 ▸78

30STRUCT+22 ▸52

22RAG+41 ▸63

89CTX+11 ▸100

4CODE+37 ▸41

32REASON+7 ▸39

25INSTR+21 ▸46

31CLASS+24 ▸55

58SUM+12 ▸70

best build per side · same suite, same graders · hover a score for cases passed

specs

75 tok/sspeed215 tok/s

—memory—

512Kcontext1.0M

$0cost / run$0.7017

—quantization—

openrouterharnessopenrouter

the lit value is the better spec — faster, smaller, cheaper, or more context · quantization & harness just describe the builds