[mle]/DaBench

MiMo-V2.6-Pro vs Muse Spark 1.3 Contributor

Side A
Model page
Side B
Model page
Overall
48
vs
Overall
66

Muse Spark 1.3 Contributor is clearly better. It wins 39 of the 45 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. Muse Spark 1.3 Contributor spends 1.5× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
482m 13s
5.9K, 89% thinking8.7K, 91% thinking
661m 42s
Agents
474m 48s
8.7K, 84% thinking14.7K, 88% thinking
573m 16s
Structured Extraction
601m 26s
3.4K, 87% thinking4.2K, 91% thinking
9037.9s
Coding
242m 58s
10.7K, 95% thinking10.8K, 95% thinking
482m 1s
Grounded Answers
621m 9s
3.2K, 89% thinking6.1K, 91% thinking
811m 1s
Instruction Following
591m 23s
4.5K, 86% thinking8.7K, 91% thinking
841m 18s
Business Logic
571m 52s
7.1K, 92% thinking6K, 95% thinking
761m 7s
Security & permissions
711m 41s
3.6K, 86% thinking8.2K, 87% thinking
761m 59s
Diagnosis & investigation
02m 54s
6.2K, 86% thinking17.2K, 90% thinking
173m 34s

What you'd run

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

SpecAB
Speed
55 tok/s
107 tok/s (better)
Machine
Hosted
Hosted
Memory at 32K
—
—
Context
1.1M (better)
1M
Cost per run
$1.80
$0.41 (better)
Cost per case
$0.012
$0.0027 (better)
Time per case
2m 13s
1m 42s (better)
Quantization
BF16
—
Harness
openrouter
openrouter

The race

3 lead changes

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

MiMo-V2.6-ProMuse Spark 1.3 Contributor
025507510050100150Perfect run4866

Every test case

153 test cases

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

6 won by MiMo-V2.6-Pro39 won by Muse Spark 1.3 Contributor108 tied