Horror: alone in an eerie place at night

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 (Reasoning)	93%
Claude Sonnet 4.6 (Reasoning)	90%
GPT-5.4 (Reasoning, Low)	90%
GPT-5.4	90%
GPT-5.4 Mini (Reasoning)	90%
GPT-5.1	89%
GPT-5	89%
Claude Sonnet 4.5	89%
Claude Sonnet 4.6	88%
GPT-5.4 Mini (Reasoning, Low)	88%
Z.AI GLM 5	88%
Z.AI GLM 5 Turbo	88%
Claude Opus 4.6 (Reasoning)	88%
Claude Opus 4.5	87%
Qwen 3.5 397B A17B	87%
Claude Sonnet 4	87%
MiniMax M2.7	87%
Qwen3 235B A22B Instruct 2507	87%
Claude Opus 4	87%
Gemini 3.1 Pro (Preview)	86%

	Score	Cost	Time
GPT-5.4 Mini (Reasoning)	90%	$0.017	22.1s
Z.AI GLM 5 Turbo	88%	$0.0072	27.5s
GPT-5.4 Nano (Reasoning, Low)	85%	$0.0045	19.4s
GPT-5.4 Mini (Reasoning, Low)	88%	$0.013	14.9s
Z.AI GLM 5	88%	$0.0075	50.7s
Qwen3 235B A22B Instruct 2507	87%	$0.0015	48.5s
GPT-5.4 Mini	86%	$0.014	15.8s
Mistral Large	86%	$0.015	24.6s
Claude 3.5 Haiku	83%	$0.0050	9.0s
GPT-5.4	90%	$0.039	1.1m
Writer: Palmyra X5	85%	$0.012	19.1s
Claude Sonnet 4.5	89%	$0.040	37.3s
Mistral Large 3	83%	$0.0037	29.1s
Qwen 3.5 397B A17B	87%	$0.012	1.3m
GPT-5.4 Nano	84%	$0.0047	16.7s
MiniMax M2.7	87%	$0.0029	1.0m
Hermes 3 405B	81%	$0.0050	26.4s
Grok 4.20 (Beta)	86%	$0.017	15.6s
GPT-5.4 (Reasoning, Low)	90%	$0.048	1.1m
Claude Sonnet 4.6	88%	$0.034	35.1s

	Score	Cost	Speed	Stability
GPT-5.4 Mini (Reasoning)	90%	$0.017	22.1s	87%
Z.AI GLM 5 Turbo	88%	$0.0072	27.5s	86%
GPT-5.4 Mini (Reasoning, Low)	88%	$0.013	14.9s	84%
GPT-5.4	90%	$0.039	1.1m	87%
GPT-5.4 (Reasoning)	93%	$0.075	2.1m	91%
Grok 4.20 (Beta)	86%	$0.017	15.6s	84%
GPT-5.4 (Reasoning, Low)	90%	$0.048	1.1m	86%
Claude Sonnet 4.5	89%	$0.040	37.3s	83%
Qwen 3.5 397B A17B	87%	$0.012	1.3m	83%
GPT-5.4 Mini	86%	$0.014	15.8s	81%
Claude Sonnet 4.6	88%	$0.034	35.1s	82%
Mistral Large	86%	$0.015	24.6s	82%
Qwen3 235B A22B Instruct 2507	87%	$0.0015	48.5s	80%
GPT-5.4 Nano	84%	$0.0047	16.7s	82%
MiniMax M2.7	87%	$0.0029	1.0m	80%
Z.AI GLM 5	88%	$0.0075	50.7s	79%
GPT-5.4 Nano (Reasoning, Low)	85%	$0.0045	19.4s	80%
Writer: Palmyra X5	85%	$0.012	19.1s	80%
Claude Sonnet 4.6 (Reasoning)	90%	$0.074	1.2m	85%
Claude Haiku 4.5	84%	$0.014	22.8s	81%

Rank	Model	Avg. Cost	Avg. Time	Stability	# 1	# 2	# 3	# 4	# 5	Total
5	GPT-5.4 (Reasoning)	$0.075	2.1m	91%	94	93	93	92	92	93%
19	Claude Sonnet 4.6 (Reasoning)	$0.074	1.2m	85%	95	92	90	88	87	90%
7	GPT-5.4 (Reasoning, Low)	$0.048	1.1m	86%	92	92	91	89	86	90%
4	GPT-5.4	$0.039	1.1m	87%	92	91	89	89	89	90%
1	GPT-5.4 Mini (Reasoning)	$0.017	22.1s	87%	92	92	91	88	86	90%
22	GPT-5.1	$0.046	2.1m	84%	94	92	89	88	85	89%
48	GPT-5	$0.061	3.9m	85%	92	91	89	88	87	89%
8	Claude Sonnet 4.5	$0.040	37.3s	83%	92	92	89	89	83	89%
11	Claude Sonnet 4.6	$0.034	35.1s	82%	94	89	87	86	85	88%
3	GPT-5.4 Mini (Reasoning, Low)	$0.013	14.9s	84%	92	89	87	87	87	88%
16	Z.AI GLM 5	$0.0075	50.7s	79%	93	93	90	88	77	88%
2	Z.AI GLM 5 Turbo	$0.0072	27.5s	86%	90	90	90	86	85	88%
32	Claude Opus 4.6 (Reasoning)	$0.086	1.2m	86%	90	89	88	87	86	88%
23	Claude Opus 4.5	$0.067	41.5s	84%	91	88	88	87	83	87%
9	Qwen 3.5 397B A17B	$0.012	1.3m	83%	89	89	86	86	85	87%
31	Claude Sonnet 4	$0.037	42.9s	78%	95	90	87	83	80	87%
15	MiniMax M2.7	$0.0029	1.0m	80%	92	89	86	84	83	87%
13	Qwen3 235B A22B Instruct 2507	$0.0015	48.5s	80%	93	88	86	85	82	87%
109	Claude Opus 4	$0.254	1.5m	83%	90	88	87	85	84	87%
90	Gemini 3.1 Pro (Preview)	$0.146	2.4m	84%	88	87	87	86	84	86%
71	Claude Opus 4.6	$0.082	1.1m	75%	96	88	87	82	77	86%
10	GPT-5.4 Mini	$0.014	15.8s	81%	90	89	87	85	80	86%
12	Mistral Large	$0.015	24.6s	82%	88	88	87	86	80	86%
6	Grok 4.20 (Beta)	$0.017	15.6s	84%	87	86	85	85	85	86%
18	Writer: Palmyra X5	$0.012	19.1s	80%	90	87	86	81	81	85%
25	Gemini 2.5 Pro	$0.035	31.3s	81%	88	88	85	83	82	85%
17	GPT-5.4 Nano (Reasoning, Low)	$0.0045	19.4s	80%	89	88	87	81	78	85%
20	Claude Haiku 4.5	$0.014	22.8s	81%	87	86	84	83	82	84%
50	Rocinante 12B	$0.0030	1.7m	74%	92	87	81	81	80	84%
76	ByteDance Seed 2.0 Mini	$0.0045	4.9m	80%	88	84	83	83	82	84%
14	GPT-5.4 Nano	$0.0047	16.7s	82%	86	84	84	83	82	84%
68	MoonshotAI: Kimi K2.5	$0.012	4.3m	80%	86	86	86	82	79	84%
64	Claude 3.5 Sonnet	$0.061	38.6s	75%	92	83	82	82	80	84%
24	Mistral Large 3	$0.0037	29.1s	78%	88	85	85	83	77	83%
40	WizardLM 2 8x22b	$0.0038	2.1m	80%	87	84	84	81	81	83%
26	GPT-5.4 Nano (Reasoning)	$0.0047	21.6s	78%	88	83	82	82	80	83%
36	DeepSeek V3 (2025-03-24)	$0.0014	20.7s	74%	90	87	83	81	74	83%
21	Mistral Small 4 (Reasoning)	$0.0022	28.7s	79%	86	85	84	82	78	83%
45	Aion 2.0	$0.0073	1.1m	77%	86	86	83	83	75	83%
44	DeepSeek V3.2	$0.0019	1.4m	77%	87	85	82	81	79	83%
27	Qwen 3.5 Flash	$0.0024	49.3s	80%	86	84	84	80	80	83%
28	Claude 3.5 Haiku	$0.0050	9.0s	78%	87	86	85	78	76	83%
59	GPT-5.2	$0.044	1.2m	78%	87	86	84	78	78	83%
33	Qwen 3.5 9B	$0.0011	1.3m	80%	84	84	83	82	80	82%
35	Grok 4.1 Fast	$0.0016	25.6s	77%	89	83	83	79	78	82%
38	Mistral Large 2	$0.015	25.6s	77%	86	84	84	82	75	82%
42	Qwen 3.5 35B	$0.013	42.6s	77%	87	82	81	80	80	82%
30	Mistral Medium 3.1	$0.0049	39.2s	80%	84	84	83	81	79	82%
34	Qwen 3 32B	$0.0016	54.7s	79%	84	83	82	81	80	82%
61	Gemini 3 Pro (Preview)	$0.053	48.7s	79%	85	84	84	79	77	82%
39	Qwen 3.5 27B	$0.011	50.3s	79%	84	83	83	82	77	82%
72	Grok 4.20 (Beta, Reasoning)	$0.028	24.1s	70%	89	86	79	77	76	81%
41	o4 Mini	$0.014	24.6s	77%	84	84	82	78	78	81%
47	Hermes 3 405B	$0.0050	26.4s	75%	86	86	86	80	69	81%
29	Ministral 3 3B	$0.0005	2.9s	78%	83	83	81	80	79	81%
43	Mistral Small Creative	$0.0009	8.8s	75%	87	85	83	75	75	81%
51	Qwen 3.5 122B	$0.014	33.9s	75%	87	81	80	80	77	81%
49	Arcee AI: Trinity Large (Preview)	$0.0000	49.3s	75%	87	83	82	76	75	81%
37	Mistral Small 4	$0.0013	11.2s	77%	85	81	81	79	77	81%
54	o4 Mini High	$0.024	41.4s	77%	84	81	80	79	78	81%
55	GPT-4o, Aug. 6th (temp=1)	$0.021	30.9s	75%	84	84	79	78	77	81%
60	Stealth: Hunter Alpha	$0.0000	1.1m	73%	86	85	80	76	75	81%
46	Qwen 3.5 Plus (2026-02-15)	$0.0076	36.0s	77%	83	83	81	80	76	80%
62	GPT-4.1	$0.018	49.1s	75%	85	82	80	78	75	80%
79	MiniMax M2.5	$0.0035	2.0m	73%	85	84	79	77	77	80%
80	Z.AI GLM 4.7	$0.0085	1.3m	71%	88	82	79	76	75	80%
53	Gemma 3 27B	$0.0009	41.6s	74%	84	83	79	76	76	80%
67	Z.AI GLM 4.5	$0.0039	23.4s	70%	88	85	81	78	67	80%
69	ByteDance Seed 2.0 Lite	$0.011	1.9m	76%	82	81	81	80	74	80%
52	Z.AI GLM 4.6	$0.0049	28.0s	75%	83	81	81	79	73	79%
56	ByteDance Seed 1.6 Flash	$0.0013	26.4s	74%	83	80	78	76	76	79%
58	LFM2 24B	$0.0003	30.9s	74%	83	81	81	76	72	78%
75	Stealth: Healer Alpha	$0.0000	19.4s	69%	89	81	79	73	71	78%
78	DeepSeek V3 (2024-12-26)	$0.0026	37.2s	70%	87	79	79	73	73	78%
73	GPT-5 Mini	$0.0093	1.1m	74%	81	81	80	75	72	78%
57	GPT-4.1 Mini	$0.0030	18.5s	74%	81	79	78	78	74	78%
65	Gemma 3 12B	$0.0008	47.9s	73%	80	80	79	79	71	78%
66	Gemini 2.5 Flash (Reasoning)	$0.010	19.0s	73%	83	79	78	76	73	78%
82	Z.AI GLM 4.7 Flash	$0.0016	1.1m	71%	82	82	77	75	72	78%
85	DeepSeek V3.1	$0.0022	1.7m	73%	80	80	78	77	71	77%
63	GPT-4o Mini (temp=1)	$0.0015	38.2s	75%	79	78	78	75	75	77%
95	Claude 3.7 Sonnet	$0.049	46.9s	73%	81	79	78	74	72	77%
84	Hermes 3 70B	$0.0016	1.5m	73%	81	80	80	73	70	77%
83	GPT-4o, May 13th (temp=1)	$0.042	13.2s	75%	78	78	78	76	73	77%
89	Cohere Command R+ (Aug. 2024)	$0.023	56.0s	73%	80	79	78	75	70	77%
70	Gemini 2.5 Flash	$0.0049	9.8s	72%	80	77	77	76	72	76%
92	DeepSeek-V2 Chat	$0.0025	45.9s	69%	83	80	75	73	71	76%
86	Ministral 8B	$0.0005	8.6s	68%	82	81	76	76	66	76%
99	ByteDance Seed 1.6	$0.0096	1.8m	69%	82	77	74	74	74	76%
88	Ministral 3 14B	$0.0011	8.3s	68%	82	80	74	73	71	76%
74	Grok 4 Fast	$0.0018	17.2s	72%	79	77	75	75	73	76%
77	Ministral 3B	$0.0002	2.5s	71%	79	79	75	74	72	76%
94	Arcee AI: Trinity Mini	$0.0004	9.4s	65%	86	76	74	72	70	76%
101	Grok 4	$0.042	1.4m	71%	80	78	75	74	72	76%
81	Gemini 3 Flash (Preview)	$0.0070	20.2s	72%	78	76	74	74	74	75%
93	Ministral 3 8B	$0.0008	6.2s	66%	82	80	75	75	65	75%
91	Gemini 3 Flash (Preview, Reasoning)	$0.011	25.7s	71%	78	77	77	75	67	75%
87	Mistral NeMO	$0.0008	7.6s	70%	79	77	76	70	68	74%
97	Gemini 2.5 Flash Lite (Reasoning)	$0.0025	21.0s	68%	77	76	72	71	69	73%
107	GPT-4o, May 13th (temp=0)	$0.058	15.0s	67%	79	74	71	71	70	73%
104	Claude 3 Haiku	$0.0028	12.3s	62%	81	79	71	66	64	72%
98	GPT-4.1 Nano	$0.0008	14.9s	67%	79	73	73	69	67	72%
100	GPT-4o Mini (temp=0)	$0.0013	40.4s	67%	76	74	71	71	68	72%
103	Llama 3.1 70B	$0.0026	48.5s	65%	81	75	75	65	63	72%
96	Gemini 2.5 Flash Lite	$0.0010	7.2s	69%	73	73	71	70	70	72%
102	Gemma 3 4B	$0.0003	21.7s	65%	79	72	71	69	66	71%
110	GPT-4o, Aug. 6th (temp=0)	$0.050	40.8s	66%	78	74	74	68	63	71%
105	Qwen 2.5 72B	$0.0013	35.7s	66%	74	73	72	67	63	70%
106	Gemini 3.1 Flash Lite (Preview)	$0.0036	8.3s	65%	75	72	72	64	64	69%
108	Nemotron 3 Super	$0.0000	51.4s	65%	75	69	69	67	66	69%
111	Llama 3.1 8B	$0.0003	1.1m	62%	77	69	68	67	65	69%
112	Llama 3.1 Nemotron 70B	$0.0061	29.7s	61%	76	70	70	66	57	68%
118	Mistral Small 3.2 24B	$0.012	10.0m	63%	71	69	67	67	61	67%
113	GPT-5 Nano	$0.0043	1.5m	62%	72	67	66	65	62	66%
114	Stealth: Aurora Alpha	$0.0000	8.5s	59%	69	64	62	62	62	64%
115	Inception Mercury 2	$0.0026	5.5s	59%	69	65	62	61	61	64%
116	Inception Mercury	$0.016	28.3s	58%	67	65	62	61	61	63%
117	Nemotron 3 Nano	$0.0010	38.8s	55%	67	65	61	58	56	61%
79.76%

Median	Evaluator	Top 3	Flop 3
80.0%	"Not X but Y" pattern overuse	100Claude Sonnet 4 100Claude 3 Haiku 100Mistral NeMO	0GPT-5 Nano 0Gemini 2.5 Flash Lite 16Gemini 2.5 Flash Lite (Reasoning)
80.0%	Adverb-first sentence starts	100LFM2 24B 100DeepSeek V3.2 100Writer: Palmyra X5	7Nemotron 3 Nano 10ByteDance Seed 2.0 Lite 16Mistral Small 3.2 24B
100.0%	Adverbs in dialogue tags	100Ministral 8B 100Gemini 3 Flash (Preview, Reasoning) 100Claude Sonnet 4.5	33GPT-4o Mini (temp=1) 50GPT-4o, May 13th (temp=1) 53Claude Sonnet 4
90.9%	AI-ism adverb frequency	100o4 Mini High 99Qwen 3.5 35B 99ByteDance Seed 2.0 Lite	68Llama 3.1 Nemotron 70B 69Claude 3 Haiku 69Claude 3.5 Haiku
100.0%	AI-ism character names	100Cohere Command R+ (Aug. 2024) 100GPT-4.1 Mini 100GPT-4.1 Nano	92Claude Opus 4 96Llama 3.1 8B 96ByteDance Seed 1.6 Flash
100.0%	AI-ism location names	100GPT-5 100Grok 4.1 Fast 100Claude Haiku 4.5	—
22.1%	AI-ism word frequency	76ByteDance Seed 2.0 Mini 70GPT-5 70GPT-5.4	0Grok 4 Fast 0Arcee AI: Trinity Mini 0Ministral 3B
100.0%	Cliché density	100Ministral 3 8B 100Grok 4 Fast 100Grok 4.20 (Beta)	47Mistral Small 3.2 24B 60Qwen 2.5 72B 67Llama 3.1 Nemotron 70B
41.7%	Dialogue tag variety (said vs. fancy)	100Claude Sonnet 4.6 100Qwen 3.5 27B 100Mistral Large	0Gemini 3.1 Flash Lite (Preview) 0Gemini 2.5 Flash 0Gemini 2.5 Flash Lite
69.9%	Em-dash & semicolon overuse	100Qwen 3.5 Flash 100Claude Sonnet 4 100Qwen 3.5 122B	0GPT-4.1 Mini 0GPT-5 Nano 0GPT-4o, Aug. 6th (temp=1)
100.0%	Emotion telling (show vs. tell)	100Gemini 3 Flash (Preview, Reasoning) 100Stealth: Hunter Alpha 100Nemotron 3 Super	77Mistral Small 3.2 24B 82GPT-4o, May 13th (temp=0) 89Inception Mercury
97.3%	Filter word density	100Qwen3 235B A22B Instruct 2507 100GPT-5.4 Mini 100Mistral Large 3	0Inception Mercury 2Llama 3.1 Nemotron 70B 11Llama 3.1 8B
100.0%	Gibberish response detection	100Gemini 3 Flash (Preview) 100Gemini 3.1 Pro (Preview) 100Llama 3.1 Nemotron 70B	48Llama 3.1 8B 80Llama 3.1 70B 88DeepSeek V3 (2025-03-24)
100.0%	Markdown formatting overuse	100Gemini 2.5 Flash (Reasoning) 100Grok 4.20 (Beta) 100GPT-5.4 (Reasoning, Low)	51ByteDance Seed 1.6 Flash 67Mistral Medium 3.1 78Mistral Small Creative
100.0%	Missing dialogue indicators (quotation marks)	100DeepSeek V3 (2025-03-24) 100Ministral 3 8B 100Z.AI GLM 5	80GPT-5 80Qwen 3.5 35B 80Llama 3.1 8B
87.5%	Name drop frequency	100Z.AI GLM 4.7 100GPT-4.1 Nano 100Claude Opus 4.6 (Reasoning)	28Rocinante 12B 51Qwen 3.5 9B 51Hermes 3 405B
73.5%	Narrator intent-glossing	100GPT-5.4 (Reasoning, Low) 100GPT-5.4 (Reasoning) 100Gemini 3.1 Pro (Preview)	0Stealth: Aurora Alpha 0Nemotron 3 Nano 1Inception Mercury
100.0%	Overuse of "that" (subordinate clause padding)	100Gemini 3 Flash (Preview) 100GPT-5.4 Nano (Reasoning) 100o4 Mini High	40ByteDance Seed 2.0 Lite 60Llama 3.1 8B 77Hermes 3 70B
100.0%	Paragraph length variance	100GPT-5.4 Nano (Reasoning) 100Claude Opus 4.6 100Mistral Small 4	16Inception Mercury 32Nemotron 3 Nano 43GPT-4o, Aug. 6th (temp=0)
98.9%	Passive voice overuse	100Claude Sonnet 4 100GPT-5.1 100Grok 4	89Inception Mercury 89Llama 3.1 Nemotron 70B 90ByteDance Seed 2.0 Mini
96.0%	Past progressive (was/were + -ing) overuse	100GPT-4o Mini (temp=0) 100LFM2 24B 100GPT-5.1	33Z.AI GLM 4.7 Flash 34Ministral 3 8B 37Gemini 3.1 Flash Lite (Preview)
92.5%	Pronoun-first sentence starts	100GPT-5.4 Mini 100Claude Opus 4.6 (Reasoning) 100Claude 3.5 Haiku	16Gemini 3.1 Flash Lite (Preview) 18ByteDance Seed 2.0 Lite 20Mistral Small 3.2 24B
96.2%	Purple prose (modifier overload)	100ByteDance Seed 2.0 Lite 100GPT-5 Nano 100Ministral 8B	79Gemini 3 Flash (Preview, Reasoning) 81Gemini 2.5 Flash (Reasoning) 84Gemini 3.1 Pro (Preview)
100.0%	Repeated phrase echo	100Gemini 3 Flash (Preview, Reasoning) 100GPT-4.1 100Qwen 3.5 9B	—
100.0%	Sentence length variance	100Qwen 3.5 27B 100Gemini 3.1 Pro (Preview) 100o4 Mini High	66Nemotron 3 Nano 73Inception Mercury 79GPT-4o, Aug. 6th (temp=0)
47.8%	Sentence opener variety	87DeepSeek V3 (2025-03-24) 85Claude 3.5 Haiku 83GPT-4o Mini (temp=1)	25GPT-5 Nano 30Qwen 3.5 35B 30Inception Mercury
52.1%	Subject-first sentence starts	100Writer: Palmyra X5 100Qwen3 235B A22B Instruct 2507 97Gemma 3 27B	0Inception Mercury 2 0Stealth: Aurora Alpha 0Qwen 3.5 27B
34.4%	Subordinate conjunction sentence starts	88Gemini 2.5 Flash Lite 75ByteDance Seed 2.0 Lite 74MiniMax M2.7	0Qwen 3.5 397B A17B 0Qwen 3.5 9B 0DeepSeek V3 (2025-03-24)
65.7%	Technical jargon density	100Qwen 3.5 35B 100Qwen 3.5 9B 100Qwen 3.5 Flash	0GPT-5 Nano 0Nemotron 3 Nano 2Gemini 2.5 Flash Lite (Reasoning)
61.1%	Useless dialogue additions	100Mistral Small 4 (Reasoning) 100GPT-5 100Qwen 3.5 397B A17B	0Gemma 3 12B 0GPT-4o, Aug. 6th (temp=0) 0Qwen 2.5 72B

Bad Writing Habits

Horror: alone in an eerie place at night

Performance Score Distribution (Top 20)

Price-Performance Score Distribution (Top 20)

Most Stable Models (Top 20)

Top Overall Models (Top 20)