[mle]/DaBench

GPT-6 Luna vs MiniMax M3.1 Flash

Side A
Model page
Side B
Model page
Overall
61
vs
Overall
35

GPT-6 Luna is clearly better. It wins 53 of the 62 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. MiniMax M3.1 Flash spends 1.2× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
6149.1s
1.9K, 73% thinking2.3K, 73% thinking
3521.3s
Agents
561m 18s
1.3K, 67% thinking3.1K, 71% thinking
3637.0s
Structured Extraction
9514.5s
999, 72% thinking1.1K, 74% thinking
487.96s
Coding
732m 18s
3.6K, 60% thinking4.6K, 66% thinking
5840.6s
Grounded Answers
6715.6s
1.1K, 72% thinking1.2K, 71% thinking
339.75s
Instruction Following
8432.3s
2K, 74% thinking1.7K, 70% thinking
5314.7s
Business Logic
3743.1s
3.2K, 93% thinking3.5K, 94% thinking
2425.9s
Security & permissions
6224.5s
1K, 68% thinking1.3K, 60% thinking
3012.7s
Diagnosis & investigation
1737.7s
1.1K, 68% thinking2.1K, 51% thinking
022.8s

What you'd run

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

SpecAB
Speed
90 tok/s
135 tok/s (better)
Machine
Hosted
Hosted
Memory at 32K
—
—
Context
1.1M (better)
1M
Cost per run
$0.10
Not measured
Cost per case
$0.00067
Not measured
Time per case
49.1s
21.3s (better)
Quantization
—
—
Harness
openrouter
openrouter

The race

3 lead changes

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

GPT-6 LunaMiniMax M3.1 Flash
025507510050100150Perfect run6135

Every test case

153 test cases

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

53 won by GPT-6 Luna9 won by MiniMax M3.1 Flash91 tied