[mle]/DaBench

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

Side A
Model page
Side B
Model page
Overall
45
vs
Overall
66

Muse Spark 1.3 Contributor is clearly better. It wins 40 of the 43 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 2.0× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
451m 21s
4.2K, 92% thinking8.7K, 91% thinking
661m 42s
Agents
381m 19s
2.2K, 86% thinking14.7K, 88% thinking
573m 16s
Structured Extraction
6238.0s
2.6K, 91% thinking4.2K, 91% thinking
9037.9s
Coding
242m 40s
9.9K, 97% thinking10.8K, 95% thinking
482m 1s
Grounded Answers
6242.1s
2.6K, 89% thinking6.1K, 91% thinking
811m 1s
Instruction Following
641m 2s
4.8K, 90% thinking8.7K, 91% thinking
841m 18s
Business Logic
4356.4s
4.4K, 96% thinking6K, 95% thinking
761m 7s
Security & permissions
662m 6s
3.8K, 88% thinking8.2K, 87% thinking
761m 59s
Diagnosis & investigation
01m 33s
1.5K, 83% 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
66 tok/s
107 tok/s (better)
Machine
Hosted
Hosted
Memory at 32K
—
—
Context
1.1M (better)
1M
Cost per run
$0.33 (better)
$0.41
Cost per case
$0.0021 (better)
$0.0027
Time per case
1m 21s (better)
1m 42s
Quantization
BF16
—
Harness
openrouter
openrouter

The race

1 lead change

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

MiMo-V2.6-FlashMuse Spark 1.3 Contributor
025507510050100150Perfect run4566

Every test case

153 test cases

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

3 won by MiMo-V2.6-Flash40 won by Muse Spark 1.3 Contributor110 tied