[mle]/DaBench

Flint vs Muse Spark 1.3 Contributor

Side A
Model page
Side B
Model page
Overall
21
vs
Overall
66

Muse Spark 1.3 Contributor is clearly better. It wins 76 of the 76 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. Flint spends 1.2× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
211m 17s
10.9K, 90% thinking8.7K, 91% thinking
661m 42s
Agents
1447.7s
4.8K, 85% thinking14.7K, 88% thinking
573m 16s
Structured Extraction
291m 17s
10.9K, 87% thinking4.2K, 91% thinking
9037.9s
Coding
141m 19s
12K, 98% thinking10.8K, 95% thinking
482m 1s
Grounded Answers
2957.7s
8.7K, 93% thinking6.1K, 91% thinking
811m 1s
Instruction Following
352m 28s
22.2K, 93% thinking8.7K, 91% thinking
841m 18s
Business Logic
101m 20s
11.7K, 78% thinking6K, 95% thinking
761m 7s
Security & permissions
3552.6s
7.1K, 90% thinking8.2K, 87% thinking
761m 59s
Diagnosis & investigation
057.0s
7.3K, 99% 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
149 tok/s (better)
107 tok/s
Machine
Hosted
Hosted
Memory at 32K
—
—
Context
256K
1M (better)
Cost per run
Not measured
$0.41
Cost per case
Not measured
$0.0027
Time per case
1m 17s (better)
1m 42s
Quantization
BF16
—
Harness
openrouter
openrouter

The race

No lead changes

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

FlintMuse Spark 1.3 Contributor
0255075100501001502166

Every test case

153 test cases

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

0 won by Flint76 won by Muse Spark 1.3 Contributor77 tied