[mle]/DaBench

Beam 1.0 501B A23B vs Mistral Large 4

Side A
Model page
Side B
Model page
Overall
57
vs
Overall
28

Beam 1.0 501B A23B is clearly better. It wins 60 of the 73 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. Mistral Large 4 spends 1.3× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
5717.2s
4K, 87% thinking6.2K, 72% thinking
281m 56s
Agents
5521.8s
4.8K, 90% thinking795, no thinking
1126.1s
Structured Extraction
819.98s
3.4K, 91% thinking288, no thinking
153.78s
Coding
4819.9s
4.5K, 70% thinking9.3K, 37% thinking
332m 48s
Grounded Answers
6911.6s
2.7K, 88% thinking8.1K, 89% thinking
382m 45s
Instruction Following
7820.9s
5.6K, 90% thinking12.9K, 84% thinking
693m 47s
Business Logic
5714.0s
4.2K, 95% thinking8.8K, 77% thinking
242m 40s
Security & permissions
5120.6s
3.2K, 88% thinking3.7K, 83% thinking
331m 9s
Diagnosis & investigation
1724.3s
3.1K, 90% thinking4.4K, 85% thinking
01m 19s

What you'd run

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

SpecAB
Speed
274 tok/s (better)
56 tok/s
Machine
Hosted
Hosted
Memory at 32K
—
—
Context
1M (better)
512K
Cost per run
$0.83 (better)
$2.67
Cost per case
$0.0054 (better)
$0.017
Time per case
17.2s (better)
1m 56s
Quantization
FP8
—
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.

Beam 1.0 501B A23BMistral Large 4
025507510050100150Perfect run5728

Every test case

153 test cases

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

60 won by Beam 1.0 501B A23B13 won by Mistral Large 480 tied