[mle]/DaBench

GLM 5.3 Flash vs Claude Haiku 5.5

Side A
Model page
Side B
Model page
Overall
35
vs
Overall
70

Claude Haiku 5.5 is clearly better. It wins 73 of the 80 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 1.2× as many thinking tokens per test case.

Thinking tokensAnswer tokens
TaskAB
Overall
3555.5s
2.8K, 89% thinking3.9K, 78% thinking
7025.5s
Agents
1037.0s
607, 81% thinking3.7K, 61% thinking
6139.5s
Structured Extraction
4847.2s
2.7K, 93% thinking2.6K, 83% thinking
9013.5s
Coding
541m 52s
5.3K, 83% thinking6.8K, 76% thinking
8630.6s
Grounded Answers
3756.8s
2.9K, 93% thinking2.2K, 78% thinking
7613.9s
Instruction Following
4340.8s
2K, 77% thinking4.3K, 82% thinking
9431.5s
Business Logic
331m 18s
5.8K, 98% thinking5.5K, 95% thinking
6726.0s
Security & permissions
5525.2s
963, 81% thinking2.7K, 70% thinking
6723.4s
Diagnosis & investigation
027.0s
563, 63% thinking2.3K, 50% thinking
1725.5s

What you'd run

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

SpecAB
Speed
60 tok/s
225 tok/s (better)
Machine
Hosted
Hosted
Memory at 32K
—
—
Context
1M
1M
Cost per run
$0.19 (better)
$0.42
Cost per case
$0.0013 (better)
$0.0028
Time per case
55.5s
25.5s (better)
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.

GLM 5.3 FlashClaude Haiku 5.5
025507510050100150Perfect run3570

Every test case

153 test cases

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

7 won by GLM 5.3 Flash73 won by Claude Haiku 5.573 tied