[mle]/DaBench

MiniMax M3.1 Flash vs Spark X2.6 4B

Side A
Model page
Side B
Model page
Overall
35
vs
Overall
15

MiniMax M3.1 Flash is clearly better. It wins 50 of the 57 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. Spark X2.6 4B spends 4.0× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
3521.3s
2.3K, 73% thinking8.2K, 84% thinking
151m 3s
Agents
3637.0s
3.1K, 71% thinking11.5K, 97% thinking
141m 32s
Structured Extraction
487.96s
1.1K, 74% thinking5.3K, 78% thinking
1039.5s
Coding
5840.6s
4.6K, 66% thinking9.1K, 48% thinking
41m 8s
Grounded Answers
339.75s
1.2K, 71% thinking6.6K, 86% thinking
1050.2s
Instruction Following
5314.7s
1.7K, 70% thinking11.9K, 90% thinking
371m 29s
Business Logic
2425.9s
3.5K, 94% thinking4.9K, 86% thinking
1036.3s
Security & permissions
3012.7s
1.3K, 60% thinking7K, 94% thinking
3154.3s
Diagnosis & investigation
022.8s
2.1K, 51% thinking11.6K, 98% thinking
01m 30s

What you'd run

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

SpecAB
Speed
135 tok/s
136 tok/s
Machine
Hosted
RTX 5090, 32 GB
Memory at 32K
—
9.4 GB
Context
1M (better)
605K
Cost per run
Not measured
$1.61 est.
Cost per case
Not measured
$0.011 est.
Time per case
21.3s (better)
1m 3s
Quantization
—
BF16
Harness
openrouter
llama.cpp

The race

No lead changes

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

MiniMax M3.1 FlashSpark X2.6 4B
0255075100501001503515

Every test case

153 test cases

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

50 won by MiniMax M3.1 Flash7 won by Spark X2.6 4B96 tied