[mle]/DaBench

Claude Haiku 5.5 vs GPT-6 Luna

Side A
Model page
Side B
Model page
Overall
70
vs
Overall
61

Claude Haiku 5.5 is clearly better. It wins 28 of the 37 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. Claude Haiku 5.5 spends 2.2× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
7025.5s
3.9K, 78% thinking1.9K, 73% thinking
6149.1s
Agents
6139.5s
3.7K, 61% thinking1.3K, 67% thinking
561m 18s
Structured Extraction
9013.5s
2.6K, 83% thinking999, 72% thinking
9514.5s
Coding
8630.6s
6.8K, 76% thinking3.6K, 60% thinking
732m 18s
Grounded Answers
7613.9s
2.2K, 78% thinking1.1K, 72% thinking
6715.6s
Instruction Following
9431.5s
4.3K, 82% thinking2K, 74% thinking
8432.3s
Business Logic
6726.0s
5.5K, 95% thinking3.2K, 93% thinking
3743.1s
Security & permissions
6723.4s
2.7K, 70% thinking1K, 68% thinking
6224.5s
Diagnosis & investigation
1725.5s
2.3K, 50% thinking1.1K, 68% thinking
1737.7s

What you'd run

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

SpecAB
Speed
225 tok/s (better)
90 tok/s
Machine
Hosted
Hosted
Memory at 32K
—
—
Context
1M
1.1M (better)
Cost per run
$0.42
$0.10 (better)
Cost per case
$0.0028
$0.00067 (better)
Time per case
25.5s (better)
49.1s
Quantization
—
—
Harness
openrouter
openrouter

The race

5 lead changes

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

Claude Haiku 5.5GPT-6 Luna
025507510050100150Perfect run7061

Every test case

153 test cases

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

28 won by Claude Haiku 5.59 won by GPT-6 Luna116 tied