[mle]/DaBench

Ling 3.1 Flash vs Spark X2.6 4B

Side A
Model page
Side B
Model page
Overall
45
vs
Overall
15

Ling 3.1 Flash is clearly better. It wins 59 of the 62 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. Spark X2.6 4B spends 1.2× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
451m 16s
7.9K, 73% thinking8.2K, 84% thinking
151m 3s
Agents
4155.9s
4.5K, 66% thinking11.5K, 97% thinking
141m 32s
Structured Extraction
711m 11s
8.1K, 69% thinking5.3K, 78% thinking
1039.5s
Coding
241m 42s
11.3K, 89% thinking9.1K, 48% thinking
41m 8s
Grounded Answers
5258.9s
6.7K, 74% thinking6.6K, 86% thinking
1050.2s
Instruction Following
761m 55s
11.4K, 77% thinking11.9K, 90% thinking
371m 29s
Business Logic
381m 22s
9.7K, 56% thinking4.9K, 86% thinking
1036.3s
Security & permissions
5456.1s
5.2K, 74% thinking7K, 94% thinking
3154.3s
Diagnosis & investigation
042.1s
3.2K, 72% thinking11.6K, 98% thinking
01m 30s

What you'd run

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

SpecAB
Speed
112 tok/s
136 tok/s
Machine
Hosted
RTX 5090, 32 GB
Memory at 32K
—
9.4 GB
Context
256K
605K (better)
Cost per run
Not measured
$1.61 est.
Cost per case
Not measured
$0.011 est.
Time per case
1m 16s
1m 3s (better)
Quantization
—
BF16
Harness
openrouter
llama.cpp

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 FlashSpark X2.6 4B
0255075100501001504515

Every test case

153 test cases

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

59 won by Ling 3.1 Flash3 won by Spark X2.6 4B91 tied