[mle]/DaBench

Muse Spark 1.3 Contributor vs Claude Haiku 5.5

Side A
Model page
Side B
Model page
Overall
66
vs
Overall
70

Too close to call. Claude Haiku 5.5 wins 19 of the 33 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 2.6× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
661m 42s
8.7K, 91% thinking3.9K, 78% thinking
7025.5s
Agents
573m 16s
14.7K, 88% thinking3.7K, 61% thinking
6139.5s
Structured Extraction
9037.9s
4.2K, 91% thinking2.6K, 83% thinking
9013.5s
Coding
482m 1s
10.8K, 95% thinking6.8K, 76% thinking
8630.6s
Grounded Answers
811m 1s
6.1K, 91% thinking2.2K, 78% thinking
7613.9s
Instruction Following
841m 18s
8.7K, 91% thinking4.3K, 82% thinking
9431.5s
Business Logic
761m 7s
6K, 95% thinking5.5K, 95% thinking
6726.0s
Security & permissions
761m 59s
8.2K, 87% thinking2.7K, 70% thinking
6723.4s
Diagnosis & investigation
173m 34s
17.2K, 90% thinking2.3K, 50% thinking
1725.5s

What you'd run

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

SpecAB
Speed
107 tok/s
225 tok/s (better)
Machine
Hosted
Hosted
Memory at 32K
—
—
Context
1M
1M
Cost per run
$0.41 (better)
$0.42
Cost per case
$0.0027 (better)
$0.0028
Time per case
1m 42s
25.5s (better)
Quantization
—
—
Harness
openrouter
openrouter

The race

4 lead changes

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

Muse Spark 1.3 ContributorClaude Haiku 5.5
025507510050100150Perfect run6670

Every test case

153 test cases

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

14 won by Muse Spark 1.3 Contributor19 won by Claude Haiku 5.5120 tied