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 (Reasoning)	93%
GPT-5.4 (Reasoning, Low)	93%
GPT-5.4	92%
GPT-5.4 Mini (Reasoning)	90%
GPT-5.4 Mini (Reasoning, Low)	88%
GPT-5.1	88%
Writer: Palmyra X5	88%
Grok 4.20 (Beta, Reasoning)	87%
Z.AI GLM 5 Turbo	87%
Qwen3 235B A22B Instruct 2507	87%
Gemini 3.1 Pro (Preview)	87%
GPT-5.4 Mini	87%
Qwen 3.5 397B A17B	86%
Rocinante 12B	86%
MiniMax M2.5	86%
Z.AI GLM 5	86%
Qwen 3.5 35B	85%
Claude Haiku 4.5	85%
GPT-5	85%
Claude Sonnet 4.5	85%

	Score	Cost	Time
Rocinante 12B	86%	$0.0008	26.2s
Qwen3 235B A22B Instruct 2507	87%	$0.0006	49.7s
GPT-5.4 Mini (Reasoning)	90%	$0.020	29.8s
Z.AI GLM 5 Turbo	87%	$0.0063	29.9s
Writer: Palmyra X5	88%	$0.010	20.5s
GPT-5.4 Mini (Reasoning, Low)	88%	$0.014	16.1s
Mistral Small 4 (Reasoning)	84%	$0.0015	21.5s
Grok 4.1 Fast	85%	$0.0014	23.5s
Mistral Small 4	83%	$0.0009	12.8s
Claude Haiku 4.5	85%	$0.0089	18.4s
MiniMax M2.5	86%	$0.0026	46.1s
Gemini 2.5 Flash (Reasoning)	83%	$0.0076	14.4s
Stealth: Hunter Alpha	84%	$0.0000	47.7s
GPT-5.4 Mini	87%	$0.012	14.1s
Qwen 3.5 35B	85%	$0.0086	33.0s
Z.AI GLM 4.7	84%	$0.0095	1.5m
Ministral 3 14B	79%	$0.0004	8.9s
Gemini 2.5 Flash Lite	83%	$0.0007	9.0s
Grok 4.20 (Beta)	84%	$0.017	17.9s
Mistral Medium 3.1	82%	$0.0041	30.6s

	Score	Consistency	Stability
GPT-5.4	92%	98%	90%
GPT-5.4 (Reasoning)	93%	95%	88%
GPT-5.4 (Reasoning, Low)	93%	96%	87%
GPT-5.4 Mini (Reasoning)	90%	96%	86%
Gemini 3.1 Pro (Preview)	87%	98%	85%
GPT-5.4 Mini (Reasoning, Low)	88%	97%	85%
GPT-5.1	88%	95%	85%
Writer: Palmyra X5	88%	97%	84%
Grok 4.1 Fast	85%	99%	84%
GPT-5	85%	98%	84%
GPT-5.4 Mini	87%	96%	83%
Qwen 3.5 35B	85%	96%	82%
Z.AI GLM 5 Turbo	87%	95%	82%
Qwen3 235B A22B Instruct 2507	87%	94%	82%
Claude Sonnet 4.5	85%	97%	82%
Claude Haiku 4.5	85%	94%	81%
Claude Opus 4.6 (Reasoning)	82%	98%	81%
Grok 4.20 (Beta)	84%	94%	81%
Claude Opus 4	85%	95%	81%
Z.AI GLM 4.7	84%	94%	80%

	Score	Cost	Speed	Stability
GPT-5.4 Mini (Reasoning)	90%	$0.020	29.8s	86%
GPT-5.4 Mini (Reasoning, Low)	88%	$0.014	16.1s	85%
Writer: Palmyra X5	88%	$0.010	20.5s	84%
Grok 4.1 Fast	85%	$0.0014	23.5s	84%
Z.AI GLM 5 Turbo	87%	$0.0063	29.9s	82%
GPT-5.4 Mini	87%	$0.012	14.1s	83%
Qwen3 235B A22B Instruct 2507	87%	$0.0006	49.7s	82%
GPT-5.4	92%	$0.045	1.4m	90%
GPT-5.4 (Reasoning, Low)	93%	$0.046	1.2m	87%
Qwen 3.5 35B	85%	$0.0086	33.0s	82%
Claude Haiku 4.5	85%	$0.0089	18.4s	81%
MiniMax M2.5	86%	$0.0026	46.1s	80%
Mistral Small 4 (Reasoning)	84%	$0.0015	21.5s	80%
Mistral Small 4	83%	$0.0009	12.8s	78%
Rocinante 12B	86%	$0.0008	26.2s	75%
Gemini 2.5 Flash Lite	83%	$0.0007	9.0s	77%
Grok 4.20 (Beta)	84%	$0.017	17.9s	81%
Z.AI GLM 5	86%	$0.0062	1.3m	78%
Claude Sonnet 4.5	85%	$0.028	34.3s	82%
GPT-5.4 Nano (Reasoning, Low)	81%	$0.0055	22.0s	79%

Rank	Model	Avg. Cost	Avg. Time	Stability	# 1	# 2	# 3	# 4	# 5	Total
62	GPT-5.4 (Reasoning)	$0.083	2.5m	88%	95	95	92	92	89	93%
9	GPT-5.4 (Reasoning, Low)	$0.046	1.2m	87%	96	94	91	91	91	93%
8	GPT-5.4	$0.045	1.4m	90%	93	93	92	91	91	92%
1	GPT-5.4 Mini (Reasoning)	$0.020	29.8s	86%	92	92	90	89	86	90%
2	GPT-5.4 Mini (Reasoning, Low)	$0.014	16.1s	85%	91	88	88	87	87	88%
47	GPT-5.1	$0.047	2.2m	85%	91	90	89	88	83	88%
3	Writer: Palmyra X5	$0.010	20.5s	84%	90	89	87	87	85	88%
26	Grok 4.20 (Beta, Reasoning)	$0.040	36.0s	80%	93	89	88	87	80	87%
5	Z.AI GLM 5 Turbo	$0.0063	29.9s	82%	90	89	87	86	82	87%
7	Qwen3 235B A22B Instruct 2507	$0.0006	49.7s	82%	90	90	87	86	81	87%
97	Gemini 3.1 Pro (Preview)	$0.105	1.7m	85%	88	87	87	86	86	87%
6	GPT-5.4 Mini	$0.012	14.1s	83%	90	87	86	85	84	87%
74	Qwen 3.5 397B A17B	$0.021	3.3m	79%	91	90	85	83	83	86%
15	Rocinante 12B	$0.0008	26.2s	75%	93	91	87	87	73	86%
12	MiniMax M2.5	$0.0026	46.1s	80%	91	88	86	85	81	86%
18	Z.AI GLM 5	$0.0062	1.3m	78%	91	88	85	85	80	86%
10	Qwen 3.5 35B	$0.0086	33.0s	82%	88	86	85	85	83	85%
11	Claude Haiku 4.5	$0.0089	18.4s	81%	88	88	86	83	81	85%
102	GPT-5	$0.066	3.3m	84%	86	86	86	85	83	85%
19	Claude Sonnet 4.5	$0.028	34.3s	82%	87	86	84	83	83	85%
4	Grok 4.1 Fast	$0.0014	23.5s	84%	85	85	85	85	84	85%
109	Claude Opus 4	$0.123	52.2s	81%	89	85	84	83	82	85%
91	Claude Opus 4.6	$0.073	1.3m	79%	89	86	85	83	78	85%
21	Qwen 3.5 122B	$0.015	41.8s	79%	88	84	83	83	82	84%
23	Qwen 3.5 9B	$0.0012	1.5m	79%	87	86	83	82	82	84%
54	Z.AI GLM 4.6	$0.0056	1.1m	71%	93	90	81	80	77	84%
36	Claude Sonnet 4.6	$0.024	36.1s	78%	90	86	85	82	78	84%
27	Aion 2.0	$0.0047	1.1m	78%	88	87	84	82	79	84%
24	Stealth: Hunter Alpha	$0.0000	47.7s	75%	92	84	84	84	75	84%
17	Grok 4.20 (Beta)	$0.017	17.9s	81%	87	86	86	81	79	84%
35	Z.AI GLM 4.7	$0.0095	1.5m	80%	87	86	85	81	80	84%
13	Mistral Small 4 (Reasoning)	$0.0015	21.5s	80%	87	86	85	82	79	84%
22	Z.AI GLM 4.7 Flash	$0.0017	1.3m	80%	86	83	83	82	82	83%
33	GPT-4.1	$0.016	51.3s	80%	85	85	83	83	79	83%
14	Mistral Small 4	$0.0009	12.8s	78%	88	85	84	79	78	83%
34	Gemini 2.5 Flash (Reasoning)	$0.0076	14.4s	74%	89	86	85	83	71	83%
43	o4 Mini High	$0.025	47.6s	80%	84	84	84	83	78	83%
16	Gemini 2.5 Flash Lite	$0.0007	9.0s	77%	87	85	81	81	79	83%
57	Qwen 3.5 27B	$0.013	1.1m	75%	89	84	81	80	79	83%
68	ByteDance Seed 2.0 Lite	$0.011	2.2m	78%	87	85	84	80	77	82%
25	MiniMax M2.7	$0.0028	1.1m	79%	86	83	83	81	80	82%
31	Qwen 3.5 Flash	$0.0017	40.7s	76%	87	84	81	81	78	82%
85	MoonshotAI: Kimi K2.5	$0.024	3.1m	80%	84	83	83	83	79	82%
90	Claude Opus 4.6 (Reasoning)	$0.071	1.2m	81%	83	83	83	82	80	82%
30	Gemma 3 27B	$0.0005	43.7s	77%	86	83	81	81	79	82%
98	Claude Opus 4.5	$0.070	1.0m	78%	85	84	81	81	79	82%
50	Gemini 2.5 Pro	$0.030	31.4s	79%	85	83	83	79	78	82%
28	Mistral Medium 3.1	$0.0041	30.6s	78%	85	84	83	81	76	82%
37	Mistral Large 2	$0.0088	23.3s	76%	85	84	81	79	77	81%
20	GPT-5.4 Nano (Reasoning, Low)	$0.0055	22.0s	79%	83	81	81	81	80	81%
32	Mistral Large	$0.0080	20.9s	77%	85	82	80	80	79	81%
59	DeepSeek V3.1	$0.0017	1.3m	74%	86	83	80	78	77	81%
39	GPT-5.4 Nano (Reasoning)	$0.0063	27.0s	76%	86	82	80	79	78	81%
67	WizardLM 2 8x22b	$0.0017	1.9m	76%	85	83	81	80	75	81%
49	GPT-5 Mini	$0.0095	1.0m	77%	83	83	81	81	76	81%
41	Mistral Small Creative	$0.0005	7.8s	72%	87	84	80	78	75	81%
42	Mistral Large 3	$0.0022	21.9s	74%	85	84	79	79	77	81%
56	ByteDance Seed 1.6 Flash	$0.0011	23.2s	71%	90	80	78	77	77	80%
44	o4 Mini	$0.014	27.2s	78%	82	82	81	79	78	80%
63	DeepSeek-V2 Chat	$0.0015	48.1s	71%	87	84	80	79	72	80%
46	Qwen 3.5 Plus (2026-02-15)	$0.0058	31.6s	75%	84	83	80	80	74	80%
70	DeepSeek V3.2	$0.0009	1.3m	73%	86	82	80	78	74	80%
100	GPT-5.2	$0.047	1.4m	75%	84	80	79	77	77	79%
29	Gemini 3.1 Flash Lite (Preview)	$0.0023	6.9s	78%	81	81	80	78	77	79%
107	Gemini 3 Pro (Preview)	$0.069	1.2m	75%	82	81	80	78	75	79%
55	DeepSeek V3 (2025-03-24)	$0.0010	1.1m	77%	82	81	81	76	76	79%
38	GPT-4.1 Nano	$0.0006	10.8s	76%	82	81	80	78	74	79%
40	Ministral 3 14B	$0.0004	8.9s	75%	82	81	81	77	73	79%
66	Gemini 2.5 Flash	$0.0037	8.5s	69%	86	83	79	79	68	79%
48	Claude 3.5 Haiku	$0.0023	9.5s	73%	85	81	80	75	74	79%
53	Gemma 3 12B	$0.0003	41.6s	75%	82	81	79	77	75	79%
83	Claude 3.5 Sonnet	$0.032	30.8s	74%	82	81	79	79	72	79%
51	Ministral 8B	$0.0002	8.6s	72%	83	80	77	76	75	78%
82	GPT-5.4 Nano	$0.0061	1.9m	75%	81	80	78	76	76	78%
84	DeepSeek V3 (2024-12-26)	$0.0018	1.4m	70%	86	78	76	76	76	78%
45	Ministral 3 3B	$0.0002	3.3s	74%	82	79	78	77	74	78%
64	Qwen 3 32B	$0.0012	47.9s	74%	82	80	78	77	74	78%
99	Claude Sonnet 4.6 (Reasoning)	$0.030	41.4s	69%	87	78	77	75	73	78%
61	GPT-4.1 Mini	$0.0024	22.4s	73%	83	78	77	75	75	78%
88	Hermes 3 70B	$0.0009	2.0m	72%	84	79	78	76	72	78%
87	Cohere Command R+ (Aug. 2024)	$0.016	45.2s	70%	83	82	77	76	71	78%
72	Grok 4 Fast	$0.0016	26.5s	71%	84	81	78	75	71	78%
75	GPT-4o, May 13th (temp=0)	$0.027	12.8s	75%	80	79	78	76	75	78%
79	GPT-4o, Aug. 6th (temp=1)	$0.017	20.8s	73%	82	80	77	75	74	78%
78	Z.AI GLM 4.5	$0.0043	47.5s	72%	84	78	77	75	74	78%
58	LFM2 24B	$0.0002	22.8s	74%	81	78	77	76	75	77%
77	Gemini 2.5 Flash Lite (Reasoning)	$0.0019	19.2s	69%	88	79	79	71	71	77%
60	Ministral 3 8B	$0.0004	8.5s	72%	83	79	77	74	74	77%
81	Stealth: Healer Alpha	$0.0000	22.1s	68%	87	81	80	69	68	77%
73	Claude 3 Haiku	$0.0020	11.8s	70%	84	76	75	75	75	77%
86	GPT-4o, Aug. 6th (temp=0)	$0.017	22.3s	70%	85	78	77	74	72	77%
65	Arcee AI: Trinity Mini	$0.0003	8.6s	71%	82	79	78	73	71	77%
52	Mistral NeMO	$0.0004	9.2s	75%	78	78	77	76	74	77%
89	Claude Sonnet 4	$0.024	36.4s	72%	80	78	77	76	72	76%
71	Ministral 3B	$0.0001	2.7s	70%	82	78	75	74	74	76%
118	Mistral Small 3.2 24B	$0.0042	6.3m	73%	80	79	78	74	71	76%
76	GPT-4o Mini (temp=1)	$0.0010	35.8s	72%	79	78	75	74	74	76%
69	Gemma 3 4B	$0.0002	21.6s	73%	80	76	75	75	75	76%
112	Grok 4	$0.046	1.7m	71%	81	76	76	74	71	76%
94	GPT-4o, May 13th (temp=1)	$0.029	15.7s	71%	81	78	77	72	70	76%
110	ByteDance Seed 1.6	$0.014	2.7m	71%	80	76	75	73	73	76%
80	Hermes 3 405B	$0.0018	23.8s	71%	79	78	76	73	70	75%
92	Llama 3.1 8B	$0.0001	34.7s	66%	86	76	75	70	69	75%
95	Gemini 3 Flash (Preview)	$0.011	29.5s	68%	83	76	75	73	69	75%
106	Claude 3.7 Sonnet	$0.031	38.7s	67%	81	78	73	73	71	75%
103	Arcee AI: Trinity Large (Preview)	$0.0000	48.5s	65%	85	76	72	71	70	75%
93	Gemini 3 Flash (Preview, Reasoning)	$0.014	34.4s	70%	78	77	76	75	67	75%
117	ByteDance Seed 2.0 Mini	$0.0043	4.8m	68%	80	79	75	73	66	75%
96	Qwen 2.5 72B	$0.0008	45.0s	68%	80	76	73	72	70	74%
101	GPT-4o Mini (temp=0)	$0.0010	36.0s	67%	79	74	72	71	67	73%
105	Llama 3.1 70B	$0.0008	25.0s	65%	79	77	72	69	65	72%
108	GPT-5 Nano	$0.0037	1.3m	66%	77	74	71	70	67	72%
114	Nemotron 3 Super	$0.0000	2.4m	68%	75	74	73	68	65	71%
111	Llama 3.1 Nemotron 70B	$0.0024	28.4s	62%	78	71	68	67	65	70%
104	Inception Mercury 2	$0.0023	5.7s	68%	71	70	70	69	68	69%
116	Nemotron 3 Nano	$0.0007	40.2s	56%	82	69	65	64	61	68%
115	Inception Mercury	$0.012	17.7s	61%	71	71	66	64	61	67%
113	Stealth: Aurora Alpha	$0.0000	8.1s	62%	70	70	66	64	62	66%
80.22%

Median	Evaluator	Top 3	Flop 3
100.0%	"Not X but Y" pattern overuse	100Ministral 3 8B 100GPT-5.4 (Reasoning, Low) 100Mistral NeMO	50DeepSeek V3.2 66Claude Opus 4.6 70Claude 3.7 Sonnet
48.6%	Adverb-first sentence starts	98Writer: Palmyra X5 97Ministral 3 14B 96Z.AI GLM 5 Turbo	0Hermes 3 405B 0Gemini 3.1 Pro (Preview) 0Nemotron 3 Nano
100.0%	Adverbs in dialogue tags	100Z.AI GLM 5 Turbo 100Llama 3.1 8B 100ByteDance Seed 1.6	60Hermes 3 405B 66GPT-4.1 Nano 67DeepSeek V3 (2025-03-24)
94.0%	AI-ism adverb frequency	100GPT-5.4 (Reasoning) 100Grok 4.1 Fast 100Qwen 3.5 9B	57Claude 3.5 Haiku 77GPT-4.1 Nano 78Claude 3 Haiku
100.0%	AI-ism character names	100Grok 4 100Z.AI GLM 4.7 100Writer: Palmyra X5	92Gemma 3 12B 92Gemma 3 4B 92Llama 3.1 8B
100.0%	AI-ism location names	100Claude Opus 4 100Gemini 2.5 Flash Lite 100Llama 3.1 70B	—
41.9%	AI-ism word frequency	86GPT-5.4 (Reasoning) 82Claude Sonnet 4.6 81Claude Sonnet 4.6 (Reasoning)	0GPT-4o, Aug. 6th (temp=0) 0Stealth: Aurora Alpha 0Gemini 2.5 Flash Lite (Reasoning)
93.3%	Cliché density	100Gemini 3 Flash (Preview) 100Claude Sonnet 4.6 100Claude Opus 4.6	13GPT-4o, May 13th (temp=0) 20Qwen 2.5 72B 33Stealth: Aurora Alpha
72.5%	Dialogue tag variety (said vs. fancy)	100ByteDance Seed 1.6 100Z.AI GLM 5 Turbo 100GPT-5.4 (Reasoning, Low)	0GPT-4o, May 13th (temp=1) 0GPT-4o Mini (temp=1) 0GPT-4o Mini (temp=0)
26.0%	Em-dash & semicolon overuse	100GPT-5.4 (Reasoning) 100GPT-5.4 Mini (Reasoning) 100Inception Mercury	0Grok 4 0ByteDance Seed 2.0 Mini 0Writer: Palmyra X5
100.0%	Emotion telling (show vs. tell)	100GPT-4o Mini (temp=1) 100Inception Mercury 100Z.AI GLM 4.7	87Llama 3.1 70B 90GPT-4.1 Mini 92Hermes 3 70B
95.4%	Filter word density	100Claude Opus 4.6 100Mistral Large 100Qwen 3.5 27B	21Claude Sonnet 4 27Stealth: Aurora Alpha 43Llama 3.1 Nemotron 70B
100.0%	Gibberish response detection	100Grok 4 100GPT-4o Mini (temp=0) 100ByteDance Seed 2.0 Mini	40Llama 3.1 8B 99ByteDance Seed 1.6 Flash 99MiniMax M2.5
100.0%	Markdown formatting overuse	100Grok 4 Fast 100Mistral NeMO 100Gemini 2.5 Flash (Reasoning)	60Ministral 3B 80Ministral 3 8B 82Ministral 3 14B
100.0%	Missing dialogue indicators (quotation marks)	100Claude Sonnet 4.6 100Grok 4.20 (Beta, Reasoning) 100DeepSeek V3 (2024-12-26)	81Grok 4 Fast 88GPT-4o, Aug. 6th (temp=1) 88Rocinante 12B
72.6%	Name drop frequency	100Claude Sonnet 4.6 (Reasoning) 100Claude Sonnet 4.6 100Gemini 2.5 Pro	9Claude 3 Haiku 20LFM2 24B 24GPT-5.4 Nano
85.8%	Narrator intent-glossing	100GPT-4.1 100GPT-5.1 100LFM2 24B	16GPT-5 Nano 22Arcee AI: Trinity Large (Preview) 27Claude Sonnet 4
100.0%	Overuse of "that" (subordinate clause padding)	100Grok 4 100Stealth: Healer Alpha 100Grok 4 Fast	76ByteDance Seed 2.0 Lite 79Llama 3.1 70B 80Gemini 2.5 Flash
100.0%	Paragraph length variance	100Qwen 3.5 122B 100Qwen 3.5 Flash 100Claude Opus 4	44Inception Mercury 47Mistral Small 3.2 24B 52Hermes 3 405B
98.0%	Passive voice overuse	100Grok 4 100Qwen 3.5 Plus (2026-02-15) 100GPT-5.1	90Claude Sonnet 4.6 90ByteDance Seed 1.6 92ByteDance Seed 2.0 Mini
97.1%	Past progressive (was/were + -ing) overuse	100GPT-5.4 Mini 100Qwen3 235B A22B Instruct 2507 100Qwen 3.5 Flash	51Stealth: Healer Alpha 52Gemini 3 Flash (Preview) 53Gemini 3 Flash (Preview, Reasoning)
88.4%	Pronoun-first sentence starts	100Mistral Large 2 100Mistral Medium 3.1 100Claude Sonnet 4.5	5Inception Mercury 33Mistral NeMO 39Qwen 3.5 35B
97.9%	Purple prose (modifier overload)	100Ministral 3 8B 100GPT-4.1 Mini 100Ministral 8B	92Gemini 3.1 Pro (Preview) 92Inception Mercury 93Qwen 3.5 Flash
100.0%	Repeated phrase echo	100DeepSeek V3.2 100Claude Opus 4 100ByteDance Seed 2.0 Mini	—
100.0%	Sentence length variance	100Gemini 3 Flash (Preview) 100Claude Sonnet 4.6 100GPT-5.4 Nano (Reasoning)	76Inception Mercury 91Stealth: Aurora Alpha 93Llama 3.1 8B
53.3%	Sentence opener variety	98Grok 4.1 Fast 93Claude 3.5 Haiku 88GPT-4o, Aug. 6th (temp=1)	26Inception Mercury 32Mistral Small 3.2 24B 35GPT-5 Nano
29.3%	Subject-first sentence starts	89Qwen3 235B A22B Instruct 2507 88Writer: Palmyra X5 88Z.AI GLM 5	0Stealth: Aurora Alpha 0Inception Mercury 2 0Inception Mercury
28.3%	Subordinate conjunction sentence starts	94Rocinante 12B 88Z.AI GLM 4.7 Flash 71Gemini 3.1 Flash Lite (Preview)	0Inception Mercury 2 0Mistral Large 2 0Mistral Small Creative
85.3%	Technical jargon density	100GPT-4o, May 13th (temp=0) 100GPT-5.4 100Qwen 3.5 397B A17B	6GPT-5 Nano 15ByteDance Seed 1.6 18ByteDance Seed 2.0 Lite
60.0%	Useless dialogue additions	100GPT-5.4 (Reasoning, Low) 100Qwen 3.5 35B 100Gemini 2.5 Flash Lite (Reasoning)	0GPT-4.1 Nano 0GPT-4.1 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)