[mle]/DaBench

Flint vs MiniMax M3.1 Flash

Side A
Model page
Side B
Model page
Overall
21
vs
Overall
35

MiniMax M3.1 Flash is clearly better. It wins 42 of the 53 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 5.8× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
211m 17s
10.9K, 90% thinking2.3K, 73% thinking
3521.3s
Agents
1447.7s
4.8K, 85% thinking3.1K, 71% thinking
3637.0s
Structured Extraction
291m 17s
10.9K, 87% thinking1.1K, 74% thinking
487.96s
Coding
141m 19s
12K, 98% thinking4.6K, 66% thinking
5840.6s
Grounded Answers
2957.7s
8.7K, 93% thinking1.2K, 71% thinking
339.75s
Instruction Following
352m 28s
22.2K, 93% thinking1.7K, 70% thinking
5314.7s
Business Logic
101m 20s
11.7K, 78% thinking3.5K, 94% thinking
2425.9s
Security & permissions
3552.6s
7.1K, 90% thinking1.3K, 60% thinking
3012.7s
Diagnosis & investigation
057.0s
7.3K, 99% 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
149 tok/s (better)
135 tok/s
Machine
Hosted
Hosted
Memory at 32K
—
—
Context
256K
1M (better)
Cost per run
Not measured
Not measured
Cost per case
Not measured
Not measured
Time per case
1m 17s
21.3s (better)
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.

FlintMiniMax M3.1 Flash
0255075100501001502135

Every test case

153 test cases

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

11 won by Flint42 won by MiniMax M3.1 Flash100 tied