Literary fiction: old friends reunite

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
Grok 4.20 (Beta, Reasoning)	91%
Mistral Large 2	88%
GPT-5.4	88%
GPT-5.4 Mini	87%
o4 Mini High	87%
GPT-5.4 (Reasoning, Low)	86%
Qwen3 235B A22B Instruct 2507	86%
Mistral Small Creative	86%
GPT-5.4 (Reasoning)	86%
Mistral Small 4	86%
Ministral 3 14B	86%
Mistral Small 4 (Reasoning)	85%
Grok 4.20 (Beta)	85%
Writer: Palmyra X5	85%
Qwen 3.5 397B A17B	85%
Mistral Large 3	85%
Mistral Medium 3.1	85%
Grok 4.1 Fast	85%
GPT-5.4 Mini (Reasoning)	84%
GPT-5.1	84%

	Score	Cost	Time
Mistral Small Creative	86%	$0.0005	8.0s
Mistral Large 2	88%	$0.0099	26.9s
Mistral Small 4	86%	$0.0012	20.4s
Ministral 3 14B	86%	$0.0005	11.5s
GPT-5.4 Mini	87%	$0.015	16.0s
Mistral Small 4 (Reasoning)	85%	$0.0017	27.6s
Grok 4.20 (Beta)	85%	$0.017	15.1s
Mistral Large 3	85%	$0.0019	19.0s
Mistral Medium 3.1	85%	$0.0039	31.4s
Qwen3 235B A22B Instruct 2507	86%	$0.0012	1.2m
LFM2 24B	82%	$0.0002	23.9s
Writer: Palmyra X5	85%	$0.011	22.5s
Grok 4.20 (Beta, Reasoning)	91%	$0.045	38.3s
Ministral 3 3B	78%	$0.0003	7.7s
Hermes 3 405B	82%	$0.0023	35.3s
GPT-5.4 Mini (Reasoning, Low)	84%	$0.015	17.3s
Llama 3.1 8B	80%	$0.0002	28.3s
Grok 4 Fast	82%	$0.0018	28.2s
Grok 4.1 Fast	85%	$0.0018	1.1m
Ministral 3 8B	80%	$0.0004	9.3s

	Score	Cost	Speed	Stability
Mistral Large 2	88%	$0.0099	26.9s	85%
Grok 4.20 (Beta, Reasoning)	91%	$0.045	38.3s	87%
GPT-5.4 Mini	87%	$0.015	16.0s	85%
Mistral Small Creative	86%	$0.0005	8.0s	82%
Mistral Small 4	86%	$0.0012	20.4s	83%
Ministral 3 14B	86%	$0.0005	11.5s	81%
Mistral Large 3	85%	$0.0019	19.0s	82%
Mistral Medium 3.1	85%	$0.0039	31.4s	82%
Writer: Palmyra X5	85%	$0.011	22.5s	82%
Mistral Small 4 (Reasoning)	85%	$0.0017	27.6s	80%
Qwen3 235B A22B Instruct 2507	86%	$0.0012	1.2m	80%
GPT-5.4 Mini (Reasoning, Low)	84%	$0.015	17.3s	80%
GPT-5.4 Mini (Reasoning)	84%	$0.025	28.5s	81%
Grok 4.1 Fast	85%	$0.0018	1.1m	80%
Grok 4.20 (Beta)	85%	$0.017	15.1s	79%
Grok 4 Fast	82%	$0.0018	28.2s	79%
GPT-5.4 Nano (Reasoning, Low)	83%	$0.0062	22.3s	79%
LFM2 24B	82%	$0.0002	23.9s	78%
Stealth: Healer Alpha	81%	$0.0000	25.3s	79%
GPT-5.4 Nano (Reasoning)	82%	$0.0069	27.2s	79%

Rank	Model	Avg. Cost	Avg. Time	Stability	# 1	# 2	# 3	# 4	# 5	Total
2	Grok 4.20 (Beta, Reasoning)	$0.045	38.3s	87%	94	92	92	92	86	91%
1	Mistral Large 2	$0.0099	26.9s	85%	91	89	89	87	83	88%
26	GPT-5.4	$0.061	1.8m	85%	89	89	88	88	85	88%
3	GPT-5.4 Mini	$0.015	16.0s	85%	88	88	87	86	84	87%
24	o4 Mini High	$0.027	51.3s	77%	94	89	85	83	82	87%
37	GPT-5.4 (Reasoning, Low)	$0.066	1.7m	84%	89	87	86	86	85	86%
11	Qwen3 235B A22B Instruct 2507	$0.0012	1.2m	80%	91	89	87	84	81	86%
4	Mistral Small Creative	$0.0005	8.0s	82%	90	89	88	83	81	86%
90	GPT-5.4 (Reasoning)	$0.097	3.2m	83%	88	86	86	86	84	86%
5	Mistral Small 4	$0.0012	20.4s	83%	88	87	86	85	83	86%
6	Ministral 3 14B	$0.0005	11.5s	81%	89	88	85	84	83	86%
10	Mistral Small 4 (Reasoning)	$0.0017	27.6s	80%	90	87	85	83	80	85%
15	Grok 4.20 (Beta)	$0.017	15.1s	79%	89	88	87	85	75	85%
9	Writer: Palmyra X5	$0.011	22.5s	82%	87	85	85	84	83	85%
27	Qwen 3.5 397B A17B	$0.011	2.2m	82%	87	86	85	83	83	85%
7	Mistral Large 3	$0.0019	19.0s	82%	87	86	85	84	82	85%
8	Mistral Medium 3.1	$0.0039	31.4s	82%	87	86	86	83	82	85%
14	Grok 4.1 Fast	$0.0018	1.1m	80%	89	85	85	82	82	85%
13	GPT-5.4 Mini (Reasoning)	$0.025	28.5s	81%	87	85	84	84	82	84%
69	GPT-5.1	$0.057	2.5m	81%	87	86	86	83	79	84%
12	GPT-5.4 Mini (Reasoning, Low)	$0.015	17.3s	80%	87	86	85	84	79	84%
89	GPT-5	$0.068	3.2m	81%	86	84	84	82	82	84%
28	Claude Sonnet 4.5	$0.032	35.6s	79%	88	85	84	82	79	84%
78	Gemini 3.1 Pro (Preview)	$0.098	1.8m	83%	84	84	83	83	83	83%
21	Z.AI GLM 5 Turbo	$0.0093	42.0s	78%	88	83	82	82	81	83%
39	Z.AI GLM 5	$0.0095	1.5m	78%	87	86	83	80	79	83%
31	Z.AI GLM 4.6	$0.0067	42.4s	75%	90	86	82	80	78	83%
109	Claude Opus 4	$0.162	1.1m	81%	85	84	84	82	80	83%
17	GPT-5.4 Nano (Reasoning, Low)	$0.0062	22.3s	79%	86	84	82	81	81	83%
59	ByteDance Seed 2.0 Lite	$0.013	2.3m	78%	87	83	82	81	81	83%
23	o4 Mini	$0.012	22.7s	77%	87	84	81	81	80	83%
61	Qwen 3.5 122B	$0.038	1.5m	78%	87	85	84	80	77	82%
22	Qwen 3.5 Flash	$0.0026	1.1m	79%	85	82	82	82	80	82%
33	Qwen 3.5 9B	$0.0009	1.2m	77%	87	82	81	81	80	82%
16	Grok 4 Fast	$0.0018	28.2s	79%	85	83	82	82	80	82%
41	Qwen 3.5 27B	$0.020	1.4m	79%	84	84	82	80	80	82%
107	MoonshotAI: Kimi K2.5	$0.022	5.0m	79%	86	83	83	80	79	82%
29	Qwen 3.5 35B	$0.017	1.1m	80%	84	82	82	82	81	82%
18	LFM2 24B	$0.0002	23.9s	78%	85	84	83	80	78	82%
25	Qwen 3.5 Plus (2026-02-15)	$0.0051	37.6s	77%	86	82	82	80	79	82%
77	Claude Opus 4.5	$0.083	1.2m	79%	84	84	83	80	78	82%
36	Hermes 3 405B	$0.0023	35.3s	75%	88	83	83	81	73	82%
20	GPT-5.4 Nano (Reasoning)	$0.0069	27.2s	79%	84	83	82	80	79	82%
51	Z.AI GLM 4.5	$0.0052	50.4s	73%	90	81	80	80	77	81%
62	MiniMax M2.7	$0.0056	57.4s	71%	89	84	80	79	74	81%
35	ByteDance Seed 1.6 Flash	$0.0015	29.4s	75%	86	85	81	79	75	81%
40	DeepSeek V3 (2025-03-24)	$0.0010	1.3m	77%	84	83	81	81	77	81%
81	Claude Sonnet 4	$0.032	48.1s	71%	88	86	80	79	72	81%
44	Qwen 3 32B	$0.0019	1.0m	76%	84	84	80	78	77	81%
19	Stealth: Healer Alpha	$0.0000	25.3s	79%	83	81	81	81	79	81%
38	Gemma 3 27B	$0.0004	51.2s	76%	84	83	81	79	76	81%
111	Claude Opus 4.6 (Reasoning)	$0.096	1.7m	75%	85	83	80	78	77	81%
45	Stealth: Hunter Alpha	$0.0000	49.1s	74%	87	82	81	79	74	81%
55	Claude 3.7 Sonnet	$0.036	42.6s	77%	83	82	81	80	76	81%
32	Ministral 3 8B	$0.0004	9.3s	74%	87	82	81	79	75	80%
108	Claude Opus 4.6	$0.083	1.4m	74%	87	81	80	77	77	80%
56	MiniMax M2.5	$0.0032	1.6m	76%	85	83	82	76	75	80%
92	Gemini 3 Pro (Preview)	$0.060	1.1m	74%	86	81	79	78	77	80%
66	WizardLM 2 8x22b	$0.0018	2.8m	79%	81	80	80	79	79	80%
50	Llama 3.1 8B	$0.0002	28.3s	73%	87	84	81	76	72	80%
63	Cohere Command R+ (Aug. 2024)	$0.014	27.6s	71%	89	84	82	73	73	80%
52	Claude Sonnet 4.6	$0.027	38.6s	77%	82	81	79	78	78	80%
75	DeepSeek-V2 Chat	$0.0018	1.2m	70%	89	79	78	77	76	80%
114	Claude Sonnet 4.6 (Reasoning)	$0.086	1.6m	74%	85	82	80	77	75	80%
100	GPT-5.2	$0.058	1.5m	74%	84	83	79	77	75	80%
76	GPT-4.1	$0.017	51.8s	71%	85	83	77	77	75	80%
34	GPT-5.4 Nano	$0.0066	24.3s	77%	82	81	80	77	77	80%
43	Gemini 3 Flash (Preview, Reasoning)	$0.013	30.8s	77%	82	82	82	77	75	79%
48	Mistral Large	$0.011	30.8s	75%	83	82	79	77	77	79%
84	Hermes 3 70B	$0.0006	1.1m	69%	89	83	79	73	73	79%
67	DeepSeek V3.2	$0.0010	2.3m	77%	81	80	80	79	76	79%
53	GPT-5 Mini	$0.0100	1.0m	77%	81	80	80	79	75	79%
30	Mistral NeMO	$0.0004	8.8s	76%	81	81	80	77	76	79%
117	ByteDance Seed 2.0 Mini	$0.0048	5.4m	73%	85	81	80	76	73	79%
58	Llama 3.1 Nemotron 70B	$0.0026	34.3s	73%	85	81	78	76	76	79%
57	Rocinante 12B	$0.0008	12.4s	71%	88	80	79	75	73	79%
64	Z.AI GLM 4.7	$0.013	1.1m	74%	83	81	79	77	74	79%
42	Gemini 3 Flash (Preview)	$0.0080	20.0s	76%	81	80	78	78	77	79%
49	Gemini 3.1 Flash Lite (Preview)	$0.0027	8.0s	73%	84	81	78	76	76	79%
47	Ministral 3 3B	$0.0003	7.7s	73%	82	82	82	78	68	78%
72	Z.AI GLM 4.7 Flash	$0.0015	1.1m	72%	84	82	78	76	72	78%
105	Grok 4	$0.051	2.0m	74%	81	81	79	77	73	78%
65	Aion 2.0	$0.0059	1.3m	75%	80	80	79	75	75	78%
46	Qwen 2.5 72B	$0.0006	33.4s	76%	79	79	78	77	76	78%
82	DeepSeek V3.1	$0.0017	1.7m	74%	81	80	78	76	74	78%
68	Gemini 2.5 Pro	$0.036	36.3s	76%	79	78	78	77	76	78%
60	Ministral 3B	$0.0001	5.4s	71%	82	81	78	76	70	77%
71	Arcee AI: Trinity Large (Preview)	$0.0000	49.0s	72%	82	79	76	75	74	77%
79	Claude 3.5 Sonnet	$0.039	34.0s	75%	79	79	78	76	74	77%
74	DeepSeek V3 (2024-12-26)	$0.0016	41.6s	71%	82	78	78	73	71	76%
103	GPT-5 Nano	$0.0042	1.5m	69%	83	81	79	71	66	76%
54	Ministral 8B	$0.0003	12.8s	74%	77	77	76	76	75	76%
86	GPT-4o, May 13th (temp=0)	$0.034	12.8s	72%	79	79	77	74	71	76%
73	GPT-4o, Aug. 6th (temp=1)	$0.017	28.8s	74%	77	77	76	75	73	76%
80	Claude Haiku 4.5	$0.010	22.0s	71%	79	77	75	73	73	76%
106	GPT-4o, May 13th (temp=1)	$0.027	14.3s	66%	82	80	74	73	67	75%
83	Gemini 2.5 Flash Lite	$0.0008	9.3s	69%	82	75	74	74	71	75%
98	GPT-4o Mini (temp=1)	$0.0011	38.5s	66%	85	75	74	72	71	75%
116	ByteDance Seed 1.6	$0.015	3.0m	70%	78	78	75	72	70	75%
70	Inception Mercury 2	$0.0022	5.5s	72%	77	75	74	74	73	74%
85	Claude 3.5 Haiku	$0.0023	8.6s	69%	79	78	75	70	70	74%
95	GPT-4.1 Nano	$0.0007	13.6s	66%	82	74	73	72	69	74%
91	Inception Mercury	$0.0095	15.4s	69%	79	76	75	75	66	74%
87	Gemini 2.5 Flash	$0.0048	10.7s	70%	78	75	74	74	68	74%
93	Gemma 3 12B	$0.0003	43.8s	69%	78	75	72	72	72	74%
94	GPT-4.1 Mini	$0.0025	19.8s	68%	81	75	75	71	67	74%
96	GPT-4o, Aug. 6th (temp=0)	$0.016	24.9s	70%	77	74	74	74	68	73%
97	Gemini 2.5 Flash Lite (Reasoning)	$0.0026	38.5s	68%	79	73	72	71	71	73%
102	Gemini 2.5 Flash (Reasoning)	$0.012	24.5s	68%	78	75	73	70	69	73%
115	Mistral Small 3.2 24B	$0.0023	1.9m	68%	77	76	74	72	65	73%
88	Gemma 3 4B	$0.0002	21.2s	70%	75	74	74	71	69	72%
99	Stealth: Aurora Alpha	$0.0000	7.6s	67%	76	76	75	71	62	72%
113	Nemotron 3 Super	$0.0000	37.2s	65%	79	70	70	70	69	72%
110	Llama 3.1 70B	$0.0009	25.7s	65%	79	72	72	71	64	72%
104	GPT-4o Mini (temp=0)	$0.0011	32.5s	68%	74	72	71	70	69	71%
101	Arcee AI: Trinity Mini	$0.0003	10.6s	68%	73	71	71	71	67	70%
112	Claude 3 Haiku	$0.0021	13.3s	66%	73	72	71	66	64	69%
118	Nemotron 3 Nano	$0.0011	1.3m	64%	69	69	68	64	62	66%
79.71%

Median	Evaluator	Top 3	Flop 3
100.0%	"Not X but Y" pattern overuse	100Llama 3.1 8B 100Gemini 3 Flash (Preview) 100Mistral Large 2	15GPT-5 Nano 40Nemotron 3 Super 65Gemini 2.5 Flash Lite (Reasoning)
39.3%	Adverb-first sentence starts	98Qwen3 235B A22B Instruct 2507 96Ministral 3 8B 96Ministral 3 14B	0Nemotron 3 Nano 0Nemotron 3 Super 0Qwen 3.5 Plus (2026-02-15)
100.0%	Adverbs in dialogue tags	100Writer: Palmyra X5 100Ministral 3 3B 100Mistral Large 3	28GPT-4.1 Nano 30Arcee AI: Trinity Large (Preview) 37GPT-4.1 Mini
87.7%	AI-ism adverb frequency	100Claude 3.5 Haiku 99GPT-5.2 99GPT-5	60GPT-4.1 Nano 64GPT-4.1 Mini 66Gemini 2.5 Flash (Reasoning)
100.0%	AI-ism character names	100GPT-5.4 Mini 100Inception Mercury 100GPT-4o, Aug. 6th (temp=0)	76Z.AI GLM 5 84MiniMax M2.7 88Claude Haiku 4.5
100.0%	AI-ism location names	100Grok 4 100Stealth: Aurora Alpha 100Claude Sonnet 4.6 (Reasoning)	96Grok 4 Fast 96Claude Opus 4 96Claude Sonnet 4
47.1%	AI-ism word frequency	87GPT-5.4 (Reasoning, Low) 86Claude Sonnet 4.6 (Reasoning) 84GPT-5.4 Mini (Reasoning)	0Claude 3.5 Haiku 0Llama 3.1 Nemotron 70B 0GPT-4o, May 13th (temp=0)
93.3%	Cliché density	100Claude Haiku 4.5 100Gemini 3 Flash (Preview) 100Gemini 3.1 Pro (Preview)	13GPT-4o Mini (temp=0) 33Claude 3 Haiku 40Mistral Small 3.2 24B
99.5%	Dialogue tag variety (said vs. fancy)	100GPT-5.4 (Reasoning, Low) 100Ministral 3 14B 100Qwen3 235B A22B Instruct 2507	21Gemma 3 4B 35GPT-4o, May 13th (temp=1) 37GPT-4o, Aug. 6th (temp=1)
61.2%	Em-dash & semicolon overuse	100Gemini 2.5 Flash (Reasoning) 100Inception Mercury 100Qwen 3.5 9B	0ByteDance Seed 1.6 0Z.AI GLM 5 1MoonshotAI: Kimi K2.5
100.0%	Emotion telling (show vs. tell)	100GPT-5.2 100ByteDance Seed 1.6 Flash 100GPT-4.1	61GPT-4o, May 13th (temp=0) 75Mistral Small 3.2 24B 86Claude 3 Haiku
100.0%	Filter word density	100GPT-5.4 Mini (Reasoning) 100Mistral Small 4 100DeepSeek V3 (2024-12-26)	35Nemotron 3 Nano 46Llama 3.1 8B 66GPT-5 Nano
100.0%	Gibberish response detection	100GPT-4o Mini (temp=0) 100LFM2 24B 100Qwen 3.5 35B	57Llama 3.1 8B 80Llama 3.1 70B 99DeepSeek V3 (2025-03-24)
100.0%	Markdown formatting overuse	100GPT-4.1 Nano 100Z.AI GLM 4.5 100Grok 4.20 (Beta, Reasoning)	80Ministral 3B 94ByteDance Seed 1.6 Flash
100.0%	Missing dialogue indicators (quotation marks)	100Grok 4 100GPT-5.4 (Reasoning, Low) 100Mistral NeMO	80Qwen 3.5 Flash 83Qwen 3.5 35B 94Qwen 3.5 9B
56.7%	Name drop frequency	100GPT-4.1 Nano 100Grok 4.20 (Beta) 100Gemma 3 4B	0GPT-5.2 0GPT-5.4 Mini (Reasoning, Low) 0GPT-5.4 Mini (Reasoning)
79.3%	Narrator intent-glossing	100Gemini 3.1 Pro (Preview) 100Grok 4.20 (Beta, Reasoning) 100Qwen 3.5 Plus (2026-02-15)	7Nemotron 3 Nano 18GPT-5.4 Nano 19Gemini 2.5 Flash Lite
100.0%	Overuse of "that" (subordinate clause padding)	100LFM2 24B 100Claude 3 Haiku 100Gemma 3 12B	65ByteDance Seed 2.0 Mini 70ByteDance Seed 1.6 80Inception Mercury
100.0%	Paragraph length variance	100MiniMax M2.7 100LFM2 24B 100Qwen 3.5 9B	53Nemotron 3 Nano 59Stealth: Aurora Alpha 61Arcee AI: Trinity Mini
99.1%	Passive voice overuse	100GPT-5.4 Nano 100Mistral Small 4 (Reasoning) 100o4 Mini	90Claude Sonnet 4.6 (Reasoning) 91ByteDance Seed 2.0 Lite 91Inception Mercury
100.0%	Past progressive (was/were + -ing) overuse	100GPT-4.1 100Llama 3.1 8B 100GPT-5.4 Mini (Reasoning, Low)	31Claude Sonnet 4.6 51Z.AI GLM 4.6 60Ministral 8B
54.5%	Pronoun-first sentence starts	100Claude Opus 4.5 100Llama 3.1 Nemotron 70B 100GPT-5.4 Mini (Reasoning, Low)	0DeepSeek V3.1 0Gemma 3 4B 0GPT-4.1 Nano
97.6%	Purple prose (modifier overload)	100GPT-5 100o4 Mini 100GPT-5.1	82GPT-4.1 Nano 88Gemini 2.5 Flash Lite 89GPT-4.1 Mini
100.0%	Repeated phrase echo	100Gemini 3 Flash (Preview) 100GPT-4o, Aug. 6th (temp=1) 100DeepSeek V3.1	—
100.0%	Sentence length variance	100GPT-5.4 (Reasoning, Low) 100Mistral Small 3.2 24B 100GPT-5.1	86Inception Mercury 89Stealth: Aurora Alpha 93Llama 3.1 70B
51.2%	Sentence opener variety	99Claude 3.5 Haiku 96Grok 4.1 Fast 87Llama 3.1 8B	33Qwen 3.5 35B 34Qwen 3.5 9B 35Qwen 3.5 122B
10.6%	Subject-first sentence starts	73Llama 3.1 8B 64Writer: Palmyra X5 63Qwen3 235B A22B Instruct 2507	0Arcee AI: Trinity Mini 0Qwen 3.5 9B 0Gemini 2.5 Flash (Reasoning)
23.2%	Subordinate conjunction sentence starts	88Grok 4 Fast 80Ministral 3 14B 73GPT-5 Nano	0GPT-5.4 Mini (Reasoning) 0Qwen 2.5 72B 0Claude Sonnet 4.6 (Reasoning)
83.5%	Technical jargon density	100Ministral 8B 100Gemma 3 27B 100Qwen 3.5 397B A17B	17ByteDance Seed 1.6 22Claude Opus 4.6 27Nemotron 3 Nano
61.1%	Useless dialogue additions	100GPT-5.4 (Reasoning) 100GPT-5.4 Nano (Reasoning) 100GPT-5.4	0GPT-4o Mini (temp=0) 0Gemma 3 4B 0DeepSeek V3 (2024-12-26)

Bad Writing Habits

Literary fiction: old friends reunite

Performance Score Distribution (Top 20)

Price-Performance Score Distribution (Top 20)

Most Stable Models (Top 20)

Top Overall Models (Top 20)