[mle]/DaBench

Clover 35B A3B Preview vs Qwen3.8 27B

Side A
Model page
Side B
Model page
Overall
30
vs
Overall
48

Qwen3.8 27B is clearly better. It wins 53 of the 67 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. The time under each score is the average per test case, on the machine it ran on. Clover 35B A3B Preview spends 1.5× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
3035.1s
8.4K, 95% thinking5.9K, 91% thinking
4833.3s
Agents
2036.8s
9.6K, 91% thinking5.2K, 78% thinking
5736.6s
Structured Extraction
4327.5s
7.2K, 97% thinking4.6K, 92% thinking
7320.6s
Coding
101m 5s
12.2K, 100% thinking9.3K, 94% thinking
3753.0s
Grounded Answers
4821.0s
5.4K, 94% thinking4.2K, 91% thinking
6131.6s
Instruction Following
5332.8s
9.2K, 95% thinking6.9K, 93% thinking
6733.6s
Business Logic
1936.1s
9.8K, 99% thinking7.6K, 98% thinking
3632.9s
Security & permissions
4725.6s
5.8K, 92% thinking3.7K, 85% thinking
5122.6s
Diagnosis & investigation
037.0s
7K, 82% thinking5.9K, 85% thinking
040.0s

What you'd run

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

SpecAB
Speed
252 tok/s (better)
196 tok/s
Machine
Hosted
Hosted
Memory at 32K
—
—
Context
—
256K
Cost per run
Not measured
$1.28
Cost per case
Not measured
$0.0083
Time per case
35.1s
33.3s (better)
Quantization
—
FP8
Harness
custom
openrouter

The race

No lead changes

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

Clover 35B A3B PreviewQwen3.8 27B
0255075100501001503048

Every test case

153 test cases

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

14 won by Clover 35B A3B Preview53 won by Qwen3.8 27B86 tied