Head to head

Any two models from the board, on one scale.

Model A

Model B

openrouter · xiaomihosted

overall

19 ▸

openrouter · deepseekhosted

41vs60

DeepSeek V4.1 Flash leads by 19 overall · ahead on 8 of 9 tasks

tasks

Task by task

9 tasks

66TOOL+12 ▸78

47STRUCT+5 ▸52

33RAG+30 ▸63

100CTXeven100

31CODE+10 ▸41

11REASON+28 ▸39

14INSTR+32 ▸46

17CLASS+38 ▸55

50SUM+20 ▸70

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

specs

69 tok/sspeed215 tok/s

—memory—

1.0Mcontext1.0M

$0.4321cost / run$0.7017

—quantization—

openrouterharnessopenrouter

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