#44

GPT-5.4

OpenAI · Release: 2026-03-05 · openai/gpt-5.4::none

Avg Score

46

Cost per result

1.496

Consistency

89

Total Cost

$0.090

Tests Correct

6

A test is fully passed only if every run passed for that test.

Wrong Tests

9

Attempt pass rate: 44.4%

Flaky tests

2

Response time: avg 1.46s · total 21.86s · max 2.89s

Wrong answer: 8 Did not follow instructions: 1

Top Models by Score

Choose the first model, then click a second model to open a side-by-side page.

#38 DeepSeek V3.2 52

#39 Qwen3.5-122B-A10B 51

#40 Qwen3.5-Flash 49

#41 Qwen3.5-27B 49

#42 MiniMax M2.5 48

#43 Qwen3.5-35B-A3B 47

#44 GPT-5.4 46

#45 Trinity Large Preview (free) 43

#46 GPT-4o-mini 41

#47 GLM 4.7 Flash 39

#48 Kimi K2.5 37

#49 Qwen3 Coder Next 36

#50 Mercury 2 34

Category	Avg Score	Consistency	Tests Correct
Anti-AI Tricks	100	73	0/3
Combined	100	100	0/1
Data parsing and extraction	99	100	2/2
Domain specific	40	72	1/3
Instructions following	55	100	1/2
Puzzle Solving	40	98	1/3
Tool Calling	100	100	1/1