[mle]/DaBench

Ling 3.1 Flash vs Mistral Large 4

Side A
Model page
Side B
Model page
Overall
45
vs
Overall
28

Ling 3.1 Flash is clearly better. It wins 43 of the 55 test cases where they differ.

Learn to build evals like this in the AI Engineering Academy

Task by task

A score turns orange only when that model wins clearly more of the task's test cases. Under each score, the average time per test case on the machine it ran on. Ling 3.1 Flash spends 1.3× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
451m 16s
7.9K, 73% thinking6.2K, 72% thinking
281m 56s
Agents
4155.9s
4.5K, 66% thinking795, no thinking
1126.1s
Structured Extraction
711m 11s
8.1K, 69% thinking288, no thinking
153.78s
Coding
241m 42s
11.3K, 89% thinking9.3K, 37% thinking
332m 48s
Grounded Answers
5258.9s
6.7K, 74% thinking8.1K, 89% thinking
382m 45s
Instruction Following
761m 55s
11.4K, 77% thinking12.9K, 84% thinking
693m 47s
Business Logic
381m 22s
9.7K, 56% thinking8.8K, 77% thinking
242m 40s
Security & permissions
5456.1s
5.2K, 74% thinking3.7K, 83% thinking
331m 9s
Diagnosis & investigation
042.1s
3.2K, 72% thinking4.4K, 85% thinking
01m 19s

What you'd run

The build behind each score. Orange marks the faster, smaller, cheaper or longer-context side.

SpecAB
Speed
112 tok/s (better)
56 tok/s
Machine
Hosted
Hosted
Memory at 32K
—
—
Context
256K
512K (better)
Cost per run
Not measured
$2.67
Cost per case
Not measured
$0.017
Time per case
1m 16s (better)
1m 56s
Quantization
—
—
Harness
openrouter
openrouter

The race

No lead changes

Each model's running score as the test cases go by. The dashed line is a perfect run.

Ling 3.1 FlashMistral Large 4
025507510050100150Perfect run4528

Every test case

153 test cases

Each square is one test case, colored by the model that scored higher.

43 won by Ling 3.1 Flash12 won by Mistral Large 498 tied