Thriller: chase through city streets

Bad Writing Habits

Detects common prose quality anti-patterns in AI-generated creative writing, including passive voice, past progressive overuse, weak dialogue tags, filter words, purple prose, cliches, AI-ism words/adverbs/names, and more.

Creative Writing Hallucination

Performance Score Distribution (Top 20)

Click a model name to view its detail page.

	Score
GPT-5.4	92%
GPT-5.4 (Reasoning)	91%
GPT-5.4 (Reasoning, Low)	91%
GPT-5.4 Mini (Reasoning)	90%
GPT-5.4 Mini	89%
GPT-5.4 Mini (Reasoning, Low)	87%
Qwen 3.5 397B A17B	87%
Gemini 3.1 Pro (Preview)	86%
Writer: Palmyra X5	86%
Qwen3 235B A22B Instruct 2507	86%
Claude Sonnet 4.5	85%
Z.AI GLM 5 Turbo	85%
Qwen 3.5 35B	84%
GPT-5.1	83%
GPT-4.1	83%
Gemini 2.5 Pro	83%
Mistral Small 4 (Reasoning)	83%
GPT-5	82%
Claude Opus 4	81%
Gemini 2.5 Flash	81%

	Score	Cost	Time
GPT-5.4 Mini (Reasoning)	90%	$0.022	26.3s
GPT-5.4 Mini	89%	$0.015	18.3s
GPT-5.4 Mini (Reasoning, Low)	87%	$0.013	16.0s
Writer: Palmyra X5	86%	$0.011	22.5s
Gemini 2.5 Flash	81%	$0.0039	8.1s
Qwen3 235B A22B Instruct 2507	86%	$0.0011	1.1m
GPT-5.4	92%	$0.050	1.4m
GPT-5.4 (Reasoning, Low)	91%	$0.051	1.4m
Mistral Small 4 (Reasoning)	83%	$0.0023	32.8s
Qwen 3.5 397B A17B	87%	$0.016	1.5m
Z.AI GLM 5 Turbo	85%	$0.0068	30.7s
GPT-4.1	83%	$0.018	46.3s
Qwen 3.5 35B	84%	$0.0068	26.9s
Mistral Small 4	79%	$0.0011	17.2s
Gemini 2.5 Flash Lite (Reasoning)	81%	$0.0027	28.7s
Ministral 8B	79%	$0.0002	8.4s
MiniMax M2.5	81%	$0.0026	35.8s
GPT-4.1 Mini	81%	$0.0025	14.4s
GPT-5.4 Nano (Reasoning, Low)	80%	$0.0049	17.8s
Claude Sonnet 4.5	85%	$0.027	34.3s

	Score	Cost	Speed	Stability
GPT-5.4 Mini	89%	$0.015	18.3s	86%
GPT-5.4 Mini (Reasoning)	90%	$0.022	26.3s	85%
GPT-5.4	92%	$0.050	1.4m	90%
GPT-5.4 Mini (Reasoning, Low)	87%	$0.013	16.0s	83%
Writer: Palmyra X5	86%	$0.011	22.5s	81%
GPT-5.4 (Reasoning, Low)	91%	$0.051	1.4m	86%
Qwen3 235B A22B Instruct 2507	86%	$0.0011	1.1m	79%
Z.AI GLM 5 Turbo	85%	$0.0068	30.7s	78%
Qwen 3.5 397B A17B	87%	$0.016	1.5m	82%
Qwen 3.5 35B	84%	$0.0068	26.9s	76%
Claude Sonnet 4.5	85%	$0.027	34.3s	78%
MiniMax M2.5	81%	$0.0026	35.8s	78%
GPT-4.1 Mini	81%	$0.0025	14.4s	77%
Gemini 2.5 Flash Lite (Reasoning)	81%	$0.0027	28.7s	78%
GPT-5.4 Nano (Reasoning, Low)	80%	$0.0049	17.8s	78%
Mistral Small 4 (Reasoning)	83%	$0.0023	32.8s	75%
GPT-4.1	83%	$0.018	46.3s	78%
Gemini 2.5 Flash (Reasoning)	81%	$0.0100	19.8s	76%
Ministral 8B	79%	$0.0002	8.4s	75%
o4 Mini High	81%	$0.022	37.6s	79%

Rank	Model	Avg. Cost	Avg. Time	Stability	# 1	# 2	# 3	# 4	# 5	Total
3	GPT-5.4	$0.050	1.4m	90%	94	92	92	91	91	92%
22	GPT-5.4 (Reasoning)	$0.086	2.5m	88%	94	93	92	89	87	91%
6	GPT-5.4 (Reasoning, Low)	$0.051	1.4m	86%	94	94	92	88	86	91%
2	GPT-5.4 Mini (Reasoning)	$0.022	26.3s	85%	94	92	92	87	85	90%
1	GPT-5.4 Mini	$0.015	18.3s	86%	92	90	89	88	86	89%
4	GPT-5.4 Mini (Reasoning, Low)	$0.013	16.0s	83%	91	88	88	88	82	87%
9	Qwen 3.5 397B A17B	$0.016	1.5m	82%	90	89	87	85	82	87%
50	Gemini 3.1 Pro (Preview)	$0.075	1.4m	80%	92	89	88	85	78	86%
5	Writer: Palmyra X5	$0.011	22.5s	81%	90	87	86	86	81	86%
7	Qwen3 235B A22B Instruct 2507	$0.0011	1.1m	79%	92	87	85	83	81	86%
11	Claude Sonnet 4.5	$0.027	34.3s	78%	91	87	84	84	81	85%
8	Z.AI GLM 5 Turbo	$0.0068	30.7s	78%	91	86	83	83	82	85%
10	Qwen 3.5 35B	$0.0068	26.9s	76%	90	86	82	81	81	84%
61	GPT-5.1	$0.060	1.5m	78%	87	86	84	80	78	83%
17	GPT-4.1	$0.018	46.3s	78%	87	86	85	81	76	83%
32	Gemini 2.5 Pro	$0.038	38.0s	77%	89	83	82	81	79	83%
16	Mistral Small 4 (Reasoning)	$0.0023	32.8s	75%	89	84	83	83	74	83%
101	GPT-5	$0.073	3.0m	80%	85	83	83	83	79	82%
115	Claude Opus 4	$0.181	1.4m	73%	87	85	80	78	76	81%
24	Gemini 2.5 Flash	$0.0039	8.1s	72%	88	86	85	79	68	81%
12	MiniMax M2.5	$0.0026	35.8s	78%	84	83	82	80	77	81%
92	Claude Opus 4.6 (Reasoning)	$0.080	1.3m	77%	85	82	81	79	78	81%
20	o4 Mini High	$0.022	37.6s	79%	82	82	81	81	79	81%
43	Hermes 3 405B	$0.0022	1.5m	75%	88	83	83	76	75	81%
49	DeepSeek V3 (2025-03-24)	$0.0012	30.8s	69%	94	81	80	76	73	81%
18	Gemini 2.5 Flash (Reasoning)	$0.0100	19.8s	76%	85	83	81	79	77	81%
13	GPT-4.1 Mini	$0.0025	14.4s	77%	86	81	80	79	79	81%
14	Gemini 2.5 Flash Lite (Reasoning)	$0.0027	28.7s	78%	83	82	81	80	77	81%
27	Mistral Medium 3.1	$0.0053	49.4s	76%	84	81	80	79	77	80%
41	DeepSeek-V2 Chat	$0.0015	33.0s	71%	88	85	79	76	75	80%
83	MoonshotAI: Kimi K2.5	$0.016	2.5m	74%	85	83	80	76	75	80%
82	Rocinante 12B	$0.0031	2.6m	73%	87	81	79	78	75	80%
34	Claude Haiku 4.5	$0.0093	19.7s	73%	86	82	80	80	72	80%
45	DeepSeek V3 (2024-12-26)	$0.0016	36.1s	71%	90	80	80	77	73	80%
15	GPT-5.4 Nano (Reasoning, Low)	$0.0049	17.8s	78%	82	82	81	77	77	80%
39	MiniMax M2.7	$0.0034	59.8s	74%	85	80	78	78	78	80%
53	Qwen 3.5 9B	$0.0007	1.0m	71%	86	82	77	77	76	80%
94	GPT-5.2	$0.054	1.5m	73%	86	81	80	79	73	80%
25	o4 Mini	$0.014	22.1s	77%	82	82	80	78	77	80%
21	Qwen 3.5 Flash	$0.0018	44.1s	77%	82	82	81	77	77	80%
26	ByteDance Seed 1.6 Flash	$0.0014	28.9s	75%	83	82	79	78	77	80%
33	Aion 2.0	$0.0049	1.1m	76%	83	81	80	79	75	80%
54	Hermes 3 70B	$0.0006	45.2s	70%	86	85	79	76	72	80%
64	Gemma 3 27B	$0.0004	50.6s	68%	88	85	77	74	74	80%
37	Claude 3.5 Haiku	$0.0034	17.3s	72%	84	84	78	78	74	80%
23	Ministral 3 8B	$0.0003	8.5s	74%	85	79	78	78	77	79%
38	Mistral Small 4	$0.0011	17.2s	71%	87	83	81	78	69	79%
42	Llama 3.1 8B	$0.0003	59.7s	74%	84	82	81	77	72	79%
19	Ministral 8B	$0.0002	8.4s	75%	83	81	80	77	75	79%
71	Grok 4.20 (Beta, Reasoning)	$0.029	23.4s	70%	86	83	77	76	74	79%
28	Gemini 2.5 Flash Lite	$0.0008	9.4s	74%	84	81	80	77	73	79%
30	GPT-5.4 Nano (Reasoning)	$0.0063	25.1s	75%	82	82	81	77	74	79%
73	Z.AI GLM 5	$0.0067	48.5s	68%	90	80	76	75	74	79%
47	Mistral Large 3	$0.0022	24.1s	71%	84	84	79	78	70	79%
66	Qwen 3.5 27B	$0.0087	47.6s	70%	85	84	77	75	74	79%
76	Claude Sonnet 4.6 (Reasoning)	$0.034	54.3s	73%	83	82	80	79	70	79%
31	Ministral 3 14B	$0.0005	13.5s	73%	83	82	77	76	75	79%
35	Grok 4.1 Fast	$0.0016	32.9s	74%	83	81	80	76	73	79%
87	Claude Opus 4.5	$0.065	57.5s	77%	80	80	79	79	75	79%
56	GPT-5 Mini	$0.011	42.4s	72%	83	82	77	76	75	79%
62	Mistral Large 2	$0.011	28.8s	70%	85	84	79	75	69	79%
95	Claude Sonnet 4.6	$0.028	37.7s	66%	90	79	76	74	73	78%
86	Claude Sonnet 4	$0.025	38.1s	68%	86	82	75	75	75	78%
40	LFM2 24B	$0.0002	29.5s	73%	84	81	78	77	73	78%
58	GPT-4o, Aug. 6th (temp=1)	$0.015	30.5s	72%	84	82	78	76	73	78%
68	DeepSeek V3.1	$0.0016	1.7m	74%	82	82	81	75	70	78%
52	Z.AI GLM 4.5	$0.0046	39.1s	73%	83	79	78	78	72	78%
78	WizardLM 2 8x22b	$0.0018	24.7s	66%	90	82	80	76	62	78%
48	Mistral Large	$0.012	33.0s	74%	81	79	78	77	75	78%
46	Mistral Small Creative	$0.0007	10.9s	71%	82	80	76	76	75	78%
29	Gemma 3 4B	$0.0002	19.0s	75%	80	80	79	76	74	78%
88	DeepSeek V3.2	$0.0011	2.8m	75%	81	80	79	75	74	78%
55	GPT-5.4 Nano	$0.0048	18.5s	71%	84	79	77	76	72	78%
51	Grok 4.20 (Beta)	$0.018	16.4s	74%	80	79	78	77	73	77%
63	Stealth: Hunter Alpha	$0.0000	58.0s	71%	84	80	79	74	71	77%
104	Claude Opus 4.6	$0.069	1.2m	73%	81	80	79	75	72	77%
93	Cohere Command R+ (Aug. 2024)	$0.021	2.1m	73%	81	78	77	76	75	77%
44	Ministral 3B	$0.0001	5.6s	72%	84	79	78	74	72	77%
109	Grok 4	$0.039	1.5m	65%	90	77	75	75	69	77%
36	Llama 3.1 Nemotron 70B	$0.0021	29.8s	75%	79	79	78	76	75	77%
72	Z.AI GLM 4.6	$0.0076	35.5s	70%	82	81	75	74	73	77%
81	Grok 4 Fast	$0.0015	24.1s	66%	84	83	75	74	67	77%
60	Mistral NeMO	$0.0005	20.5s	70%	83	77	75	75	73	77%
57	Gemma 3 12B	$0.0002	37.8s	72%	80	78	75	75	75	77%
79	GPT-4.1 Nano	$0.0006	10.0s	66%	87	81	76	70	68	76%
102	Claude 3.5 Sonnet	$0.042	38.5s	69%	83	79	74	74	73	76%
70	Qwen 3.5 122B	$0.012	45.0s	72%	80	78	77	75	71	76%
89	Claude 3.7 Sonnet	$0.037	46.5s	73%	78	78	77	77	71	76%
105	ByteDance Seed 1.6	$0.013	2.6m	71%	80	79	77	75	69	76%
74	Qwen 2.5 72B	$0.0008	39.1s	69%	82	78	75	72	72	76%
85	Claude 3 Haiku	$0.0021	15.8s	66%	84	79	74	73	68	76%
75	Arcee AI: Trinity Large (Preview)	$0.0000	38.1s	69%	81	79	76	74	67	76%
67	GPT-4o Mini (temp=1)	$0.0011	23.6s	71%	79	79	76	72	71	76%
59	Qwen 3.5 Plus (2026-02-15)	$0.0049	25.3s	73%	78	77	75	74	74	75%
77	GPT-4o, May 13th (temp=1)	$0.028	14.2s	72%	79	76	75	74	73	75%
69	Qwen 3 32B	$0.0012	50.5s	73%	77	77	75	74	73	75%
84	Z.AI GLM 4.7 Flash	$0.0015	1.1m	70%	80	77	76	71	71	75%
80	Llama 3.1 70B	$0.0007	23.3s	68%	81	79	79	74	62	75%
91	GPT-4o, May 13th (temp=0)	$0.026	13.3s	70%	79	75	73	73	73	75%
65	Arcee AI: Trinity Mini	$0.0003	10.0s	71%	78	76	74	73	71	74%
90	Ministral 3 3B	$0.0003	6.5s	65%	84	75	72	71	70	74%
97	GPT-4o, Aug. 6th (temp=0)	$0.016	21.9s	67%	81	76	73	71	71	74%
96	Stealth: Healer Alpha	$0.0000	22.3s	65%	80	77	73	72	64	74%
108	Z.AI GLM 4.7	$0.0098	2.0m	68%	77	73	71	71	70	72%
98	GPT-4o Mini (temp=0)	$0.0011	26.3s	68%	76	72	71	70	69	72%
100	Gemini 3 Flash (Preview, Reasoning)	$0.010	24.3s	69%	73	72	71	71	70	71%
99	Gemini 3 Flash (Preview)	$0.0066	17.3s	68%	75	72	72	68	68	71%
116	ByteDance Seed 2.0 Lite	$0.012	2.7m	57%	80	78	66	66	65	71%
110	Gemini 3 Pro (Preview)	$0.049	49.2s	66%	75	72	70	69	67	71%
103	Nemotron 3 Super	$0.0000	56.8s	69%	72	72	71	69	68	71%
107	Gemini 3.1 Flash Lite (Preview)	$0.0025	7.6s	62%	80	72	70	65	64	70%
117	ByteDance Seed 2.0 Mini	$0.0047	5.3m	63%	75	74	68	67	66	70%
106	Inception Mercury	$0.010	14.7s	67%	71	71	71	66	64	69%
118	Mistral Small 3.2 24B	$0.011	7.1m	67%	71	69	69	68	66	69%
113	GPT-5 Nano	$0.0044	1.6m	64%	68	68	68	66	61	66%
112	Nemotron 3 Nano	$0.0006	1.1m	62%	69	67	65	64	62	66%
111	Inception Mercury 2	$0.0030	8.2s	62%	65	64	63	63	62	63%
114	Stealth: Aurora Alpha	$0.0000	14.1s	60%	67	64	62	62	60	63%
78.26%

Median	Evaluator	Top 3	Flop 3
98.3%	"Not X but Y" pattern overuse	100Claude Opus 4.6 (Reasoning) 100Mistral Large 2 100GPT-4.1 Mini	7GPT-5 Nano 36Mistral Small 4 40Stealth: Healer Alpha
57.3%	Adverb-first sentence starts	100Writer: Palmyra X5 100Qwen3 235B A22B Instruct 2507 100GPT-5.4 (Reasoning, Low)	0Inception Mercury 2 0Stealth: Aurora Alpha 2Mistral Small 3.2 24B
95.0%	Adverbs in dialogue tags	100Stealth: Hunter Alpha 100Nemotron 3 Super 100Mistral Small Creative	20GPT-5 Nano 31Claude Opus 4 40DeepSeek V3 (2025-03-24)
91.5%	AI-ism adverb frequency	99ByteDance Seed 2.0 Lite 99o4 Mini 98Grok 4.1 Fast	81Claude Sonnet 4.6 82Gemma 3 4B 82Arcee AI: Trinity Large (Preview)
100.0%	AI-ism character names	100Claude 3.7 Sonnet 100Grok 4.1 Fast 100Qwen 3.5 397B A17B	76Llama 3.1 8B 92Claude Opus 4 96Claude Haiku 4.5
100.0%	AI-ism location names	100Qwen 3.5 27B 100Mistral Large 100Llama 3.1 Nemotron 70B	—
43.7%	AI-ism word frequency	83Claude Sonnet 4.6 82ByteDance Seed 2.0 Lite 78Claude Opus 4.6 (Reasoning)	0Mistral Small 3.2 24B 0Stealth: Aurora Alpha 0GPT-4o Mini (temp=0)
100.0%	Cliché density	100MiniMax M2.5 100Mistral Small 4 (Reasoning) 100Grok 4.1 Fast	33Qwen 2.5 72B 33GPT-4o, May 13th (temp=0) 33Claude 3 Haiku
59.1%	Dialogue tag variety (said vs. fancy)	100Gemini 2.5 Flash Lite (Reasoning) 100MiniMax M2.7 100Claude Sonnet 4.6 (Reasoning)	0Gemini 3 Flash (Preview, Reasoning) 0Gemini 3 Pro (Preview) 1Z.AI GLM 4.7
1.0%	Em-dash & semicolon overuse	100WizardLM 2 8x22b 100Inception Mercury 100Mistral NeMO	0Mistral Medium 3.1 0ByteDance Seed 1.6 0ByteDance Seed 1.6 Flash
100.0%	Emotion telling (show vs. tell)	100Gemini 2.5 Flash Lite 100Mistral Large 3 100Rocinante 12B	80WizardLM 2 8x22b 90Claude 3 Haiku 95GPT-4o Mini (temp=1)
91.0%	Filter word density	100Rocinante 12B 100Gemini 3.1 Pro (Preview) 100Mistral Small 4	0Nemotron 3 Nano 1Inception Mercury 2 2Stealth: Aurora Alpha
100.0%	Gibberish response detection	100Stealth: Hunter Alpha 100Inception Mercury 2 100Qwen 2.5 72B	0Llama 3.1 8B 35WizardLM 2 8x22b 68Rocinante 12B
100.0%	Markdown formatting overuse	100Claude Opus 4 100Grok 4.20 (Beta, Reasoning) 100Gemini 3.1 Pro (Preview)	80Ministral 3 14B 80Ministral 3 3B 80Rocinante 12B
100.0%	Missing dialogue indicators (quotation marks)	100Claude Opus 4.6 (Reasoning) 100GPT-4.1 Nano 100Qwen 3.5 Plus (2026-02-15)	60Qwen 3.5 397B A17B 80Gemini 2.5 Flash Lite (Reasoning) 80Qwen 3.5 Flash
82.6%	Name drop frequency	100Gemini 2.5 Flash (Reasoning) 100Qwen 3.5 Flash 100Claude Sonnet 4.6	18Claude 3.7 Sonnet 27ByteDance Seed 1.6 Flash 37Hermes 3 405B
74.9%	Narrator intent-glossing	100Writer: Palmyra X5 100Ministral 3B 100Ministral 3 3B	0Nemotron 3 Nano 1Inception Mercury 2 9Stealth: Aurora Alpha
100.0%	Overuse of "that" (subordinate clause padding)	100Grok 4 100DeepSeek V3.2 100Writer: Palmyra X5	50ByteDance Seed 2.0 Lite 62Claude Sonnet 4.6 67ByteDance Seed 2.0 Mini
97.1%	Paragraph length variance	100Aion 2.0 100GPT-5.4 Mini 100Qwen 3 32B	38GPT-5 Nano 43ByteDance Seed 2.0 Lite 44Llama 3.1 70B
94.8%	Passive voice overuse	100o4 Mini High 100Grok 4.1 Fast 99Grok 4 Fast	83ByteDance Seed 2.0 Lite 85ByteDance Seed 2.0 Mini 86Gemini 3 Flash (Preview, Reasoning)
84.2%	Past progressive (was/were + -ing) overuse	100GPT-5 Nano 100Gemini 2.5 Flash Lite (Reasoning) 100GPT-4.1 Mini	17Gemma 3 27B 19Z.AI GLM 4.7 23Qwen 3.5 122B
84.1%	Pronoun-first sentence starts	100Llama 3.1 8B 100Z.AI GLM 4.5 100Claude 3.7 Sonnet	9Inception Mercury 11Mistral Small 3.2 24B 23Qwen 3.5 122B
97.6%	Purple prose (modifier overload)	100Gemini 3 Flash (Preview) 100Mistral Large 3 100DeepSeek-V2 Chat	90Llama 3.1 8B 90GPT-5 Nano 91Gemini 3.1 Pro (Preview)
100.0%	Repeated phrase echo	100Gemini 3 Pro (Preview) 100DeepSeek V3.1 100Z.AI GLM 4.6	—
100.0%	Sentence length variance	100Mistral Large 3 100Gemini 2.5 Flash Lite 100Rocinante 12B	82Llama 3.1 70B 82Inception Mercury 92GPT-4o, Aug. 6th (temp=0)
49.9%	Sentence opener variety	100Llama 3.1 8B 88Claude 3.5 Haiku 86Grok 4.1 Fast	27GPT-5 Nano 27Mistral Small 3.2 24B 28Inception Mercury
48.9%	Subject-first sentence starts	100Llama 3.1 8B 96GPT-5.4 (Reasoning) 94GPT-5.4	0Inception Mercury 2 0Stealth: Aurora Alpha 1Inception Mercury
51.4%	Subordinate conjunction sentence starts	100Qwen 3.5 397B A17B 96Qwen 3.5 35B 95GPT-5.4 (Reasoning)	0Mistral Large 3 0Inception Mercury 0Grok 4
67.7%	Technical jargon density	100Aion 2.0 98GPT-4o Mini (temp=0) 96DeepSeek-V2 Chat	0MiniMax M2.5 0GPT-5 Nano 0Nemotron 3 Super
46.4%	Useless dialogue additions	100WizardLM 2 8x22b 100Gemini 2.5 Flash Lite (Reasoning) 100GPT-5.4 Mini (Reasoning)	0GPT-4o, May 13th (temp=0) 0Arcee AI: Trinity Mini 0Gemma 3 4B

Bad Writing Habits

Thriller: chase through city streets

Performance Score Distribution (Top 20)

Price-Performance Score Distribution (Top 20)

Most Stable Models (Top 20)

Top Overall Models (Top 20)