Head to head

Any two models from the board, on one scale.

Model A

Model B

Ling-3.0 flash

openrouter · inclusionaihosted

overall

21 ▸

DeepSeek V4.1 Flash

openrouter · deepseekhosted

39vs60

DeepSeek V4.1 Flash leads by 21 overall · ahead on 9 of 9 tasks

Learn to build evals like this — AI Engineering Academy →

the tape

Every case, one strip

238 cases

Ling-3.0 flash takes 0DeepSeek V4.1 Flash takes 00 even238 ran on one side only

one cell per case, in suite order · the brighter score wins the cell · hover for the case

tasks

Task by task

9 tasks

67TOOL+11 ▸78

37STRUCT+15 ▸52

30RAG+33 ▸63

89CTX+11 ▸100

11CODE+30 ▸41

32REASON+7 ▸39

7INSTR+39 ▸46

45CLASS+10 ▸55

34SUM+36 ▸70

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

tokens

Thinking spend

Ling-3.0 flash

4.1K tok/case · 55% thinking

DeepSeek V4.1 Flash

not run

6.7KTOOLnot run—

3.0KSTRUCTnot run—

3.3KRAGnot run—

874CTXnot run—

8.0KCODEnot run—

6.2KREASONnot run—

3.2KINSTRnot run—

2.8KCLASSnot run—

2.0KSUMnot run—

Only Ling-3.0 flash shows visible thinking on this suite

bar = average output tokens per case, shared scale · solid = answer · faded = thinking · hover a bar for the split

specs

What you'd run

264 tok/sspeed215 tok/s

—memory—

262.1Kcontext1.0M

$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