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
GPT-5.4	89%
GPT-5.4 (Reasoning, Low)	89%
GPT-5.4 (Reasoning)	88%
GPT-5.4 Mini	88%
GPT-5.4 Mini (Reasoning, Low)	87%
GPT-5.4 Mini (Reasoning)	86%
Mistral Large	85%
DeepSeek V3 (2025-03-24)	85%
GPT-5.1	84%
Mistral Small Creative	83%
Mistral Medium 3.1	83%
Grok 4.20 (Beta)	83%
Qwen3 235B A22B Instruct 2507	83%
Grok 4.20 (Beta, Reasoning)	83%
Writer: Palmyra X5	83%
Ministral 8B	83%
ByteDance Seed 1.6 Flash	83%
Ministral 3 14B	82%
Mistral Large 2	82%
Claude Opus 4	82%

	Score	Cost	Time
GPT-5.4 Mini	88%	$0.018	20.0s
DeepSeek V3 (2025-03-24)	85%	$0.0012	39.8s
Mistral Small Creative	83%	$0.0007	10.0s
GPT-5.4 Mini (Reasoning, Low)	87%	$0.018	21.3s
Ministral 8B	83%	$0.0003	12.7s
Ministral 3 14B	82%	$0.0006	17.7s
Mistral Large	85%	$0.012	33.8s
GPT-5.4 Mini (Reasoning)	86%	$0.030	33.3s
Writer: Palmyra X5	83%	$0.011	23.1s
Mistral Medium 3.1	83%	$0.0047	41.0s
ByteDance Seed 1.6 Flash	83%	$0.0013	30.7s
Grok 4.20 (Beta)	83%	$0.016	14.2s
Ministral 3B	78%	$0.0001	7.0s
o4 Mini	81%	$0.017	27.0s
Mistral Small 4	81%	$0.0015	23.5s
Mistral Large 3	79%	$0.0028	31.2s
Grok 4.1 Fast	81%	$0.0019	38.6s
Qwen3 235B A22B Instruct 2507	83%	$0.0011	1.2m
GPT-5.4 Nano (Reasoning, Low)	78%	$0.0054	19.1s
Mistral Small 3.2 24B	78%	$0.0006	22.9s

	Score	Consistency	Stability
GPT-5.4 (Reasoning)	88%	98%	86%
GPT-5.4 Mini	88%	96%	86%
GPT-5.4	89%	96%	85%
GPT-5.4 Mini (Reasoning, Low)	87%	97%	85%
GPT-5.4 (Reasoning, Low)	89%	95%	84%
GPT-5.4 Mini (Reasoning)	86%	96%	84%
GPT-5.1	84%	97%	83%
Grok 4.1 Fast	81%	98%	80%
Grok 4.20 (Beta, Reasoning)	83%	96%	80%
Ministral 3 14B	82%	96%	79%
Mistral Medium 3.1	83%	96%	79%
Grok 4	81%	97%	79%
Mistral Small Creative	83%	93%	79%
DeepSeek V3 (2025-03-24)	85%	91%	79%
Mistral Large	85%	92%	78%
Ministral 8B	83%	93%	78%
Qwen 3 32B	81%	97%	78%
GPT-5	81%	94%	78%
o4 Mini	81%	94%	78%
Grok 4.20 (Beta)	83%	95%	78%

	Score	Cost	Speed	Stability
GPT-5.4 Mini	88%	$0.018	20.0s	86%
GPT-5.4 Mini (Reasoning, Low)	87%	$0.018	21.3s	85%
GPT-5.4 Mini (Reasoning)	86%	$0.030	33.3s	84%
Mistral Small Creative	83%	$0.0007	10.0s	79%
Ministral 3 14B	82%	$0.0006	17.7s	79%
DeepSeek V3 (2025-03-24)	85%	$0.0012	39.8s	79%
Ministral 8B	83%	$0.0003	12.7s	78%
Mistral Large	85%	$0.012	33.8s	78%
Mistral Medium 3.1	83%	$0.0047	41.0s	79%
Grok 4.20 (Beta)	83%	$0.016	14.2s	78%
Grok 4.1 Fast	81%	$0.0019	38.6s	80%
Writer: Palmyra X5	83%	$0.011	23.1s	77%
ByteDance Seed 1.6 Flash	83%	$0.0013	30.7s	77%
Mistral Small 4	81%	$0.0015	23.5s	75%
GPT-5.4	89%	$0.070	2.0m	85%
o4 Mini	81%	$0.017	27.0s	78%
Qwen3 235B A22B Instruct 2507	83%	$0.0011	1.2m	78%
GPT-5.4 (Reasoning, Low)	89%	$0.070	1.9m	84%
GPT-4.1	81%	$0.017	34.0s	77%
Mistral Small 4 (Reasoning)	81%	$0.0024	37.2s	75%

Rank	Model	Avg. Cost	Avg. Time	Stability	# 1	# 2	# 3	# 4	# 5	Total
15	GPT-5.4	$0.070	2.0m	85%	92	91	89	88	87	89%
18	GPT-5.4 (Reasoning, Low)	$0.070	1.9m	84%	92	91	88	87	86	89%
81	GPT-5.4 (Reasoning)	$0.125	3.6m	86%	89	89	88	88	87	88%
1	GPT-5.4 Mini	$0.018	20.0s	86%	90	89	89	86	85	88%
2	GPT-5.4 Mini (Reasoning, Low)	$0.018	21.3s	85%	90	88	87	86	86	87%
3	GPT-5.4 Mini (Reasoning)	$0.030	33.3s	84%	89	87	87	85	83	86%
8	Mistral Large	$0.012	33.8s	78%	90	89	85	84	79	85%
6	DeepSeek V3 (2025-03-24)	$0.0012	39.8s	79%	91	87	86	80	78	85%
39	GPT-5.1	$0.066	1.5m	83%	85	85	85	83	82	84%
4	Mistral Small Creative	$0.0007	10.0s	79%	87	85	85	84	77	83%
9	Mistral Medium 3.1	$0.0047	41.0s	79%	88	83	83	82	81	83%
10	Grok 4.20 (Beta)	$0.016	14.2s	78%	88	84	82	82	81	83%
17	Qwen3 235B A22B Instruct 2507	$0.0011	1.2m	78%	88	85	83	82	79	83%
22	Grok 4.20 (Beta, Reasoning)	$0.052	43.2s	80%	86	85	83	81	80	83%
12	Writer: Palmyra X5	$0.011	23.1s	77%	89	85	84	79	79	83%
7	Ministral 8B	$0.0003	12.7s	78%	87	84	83	81	77	83%
13	ByteDance Seed 1.6 Flash	$0.0013	30.7s	77%	88	83	82	81	78	83%
5	Ministral 3 14B	$0.0006	17.7s	79%	85	84	83	82	79	82%
23	Mistral Large 2	$0.013	34.9s	74%	89	85	80	79	78	82%
112	Claude Opus 4	$0.190	1.4m	77%	86	84	83	78	77	82%
11	Grok 4.1 Fast	$0.0019	38.6s	80%	83	82	81	81	80	81%
19	GPT-4.1	$0.017	34.0s	77%	86	81	81	81	78	81%
100	GPT-5	$0.074	2.9m	78%	84	84	82	79	78	81%
14	Mistral Small 4	$0.0015	23.5s	75%	88	82	81	79	78	81%
25	Z.AI GLM 5 Turbo	$0.0093	40.2s	75%	87	85	82	79	75	81%
16	o4 Mini	$0.017	27.0s	78%	84	83	83	81	76	81%
20	Mistral Small 4 (Reasoning)	$0.0024	37.2s	75%	87	82	80	79	77	81%
71	MoonshotAI: Kimi K2.5	$0.016	2.6m	75%	85	84	82	81	74	81%
58	Grok 4	$0.047	1.8m	79%	82	82	81	81	78	81%
40	Claude Sonnet 4.5	$0.035	41.2s	77%	86	82	82	78	77	81%
27	Qwen 3 32B	$0.0015	1.4m	78%	83	81	81	80	78	81%
32	o4 Mini High	$0.027	43.2s	77%	84	81	80	80	78	81%
50	Claude Sonnet 4	$0.035	53.9s	75%	85	81	79	78	78	80%
28	Ministral 3 8B	$0.0004	13.8s	71%	87	83	78	77	76	80%
90	Gemini 3.1 Pro (Preview)	$0.073	1.3m	72%	89	82	80	76	74	80%
34	Qwen 3.5 9B	$0.0010	1.2m	77%	82	81	80	79	77	80%
37	MiniMax M2.7	$0.0036	1.4m	77%	81	81	80	79	77	80%
26	GPT-5.4 Nano (Reasoning)	$0.0063	24.9s	74%	85	81	79	78	76	80%
47	Claude 3.5 Sonnet	$0.048	41.3s	77%	83	81	81	77	76	80%
38	Qwen 3.5 Flash	$0.0028	1.2m	76%	82	80	79	79	78	80%
42	MiniMax M2.5	$0.0041	1.1m	75%	84	81	80	78	75	80%
41	Mistral Large 3	$0.0028	31.2s	72%	86	84	81	73	72	79%
43	Rocinante 12B	$0.0010	31.8s	71%	86	83	78	75	74	79%
35	Claude 3.5 Haiku	$0.0035	17.2s	72%	84	83	78	76	75	79%
44	Qwen 3.5 27B	$0.012	1.1m	75%	83	79	78	77	77	79%
101	Claude Opus 4.6	$0.081	1.3m	72%	83	83	80	80	69	79%
55	Arcee AI: Trinity Large (Preview)	$0.0000	32.8s	68%	90	81	79	73	71	79%
31	GPT-4o, May 13th (temp=0)	$0.026	13.1s	76%	81	80	80	78	75	79%
48	DeepSeek V3 (2024-12-26)	$0.0019	58.2s	73%	82	82	79	76	72	78%
21	GPT-5.4 Nano (Reasoning, Low)	$0.0054	19.1s	77%	80	79	79	76	76	78%
30	Claude Haiku 4.5	$0.010	21.0s	75%	82	79	79	78	74	78%
24	Grok 4 Fast	$0.0017	30.7s	76%	80	79	79	78	76	78%
111	Claude Opus 4.6 (Reasoning)	$0.112	1.8m	74%	81	80	78	78	74	78%
77	Qwen 3.5 397B A17B	$0.020	2.0m	74%	82	79	79	78	73	78%
59	Gemini 2.5 Pro	$0.043	44.1s	75%	81	80	79	77	73	78%
60	Hermes 3 405B	$0.0020	1.6m	73%	83	80	78	76	73	78%
104	GPT-5.2	$0.076	1.9m	76%	80	79	79	75	75	78%
65	Claude Sonnet 4.6	$0.035	45.3s	72%	82	80	77	77	73	78%
36	Mistral NeMO	$0.0004	19.7s	73%	81	79	78	77	73	78%
33	Mistral Small 3.2 24B	$0.0006	22.9s	75%	80	79	79	78	72	78%
45	Qwen 2.5 72B	$0.0008	33.9s	72%	83	80	78	73	73	78%
29	Ministral 3B	$0.0001	7.0s	73%	80	79	79	79	70	78%
116	ByteDance Seed 2.0 Mini	$0.0044	4.9m	70%	84	81	78	74	70	77%
61	Aion 2.0	$0.0063	1.5m	74%	81	78	78	75	74	77%
52	Llama 3.1 Nemotron 70B	$0.0026	41.8s	72%	81	79	76	75	73	77%
54	Gemini 2.5 Flash Lite	$0.0009	11.6s	68%	83	82	77	75	67	77%
103	ByteDance Seed 1.6	$0.015	2.9m	74%	79	77	76	75	75	77%
46	Stealth: Healer Alpha	$0.0000	24.2s	72%	81	79	77	73	72	76%
85	Z.AI GLM 4.7	$0.010	1.7m	71%	81	78	75	75	72	76%
78	Llama 3.1 8B	$0.0002	41.3s	65%	86	80	73	72	70	76%
49	Qwen 3.5 122B	$0.015	36.9s	75%	77	77	77	75	74	76%
82	DeepSeek V3.2	$0.0013	1.1m	67%	82	82	74	73	70	76%
56	GPT-5 Mini	$0.011	43.2s	73%	78	78	76	76	73	76%
102	Claude Opus 4.5	$0.077	1.0m	73%	80	79	78	72	72	76%
64	Gemma 3 27B	$0.0004	1.1m	71%	80	79	76	74	72	76%
99	Z.AI GLM 5	$0.0081	1.5m	66%	86	76	74	72	72	76%
53	GPT-5.4 Nano	$0.0057	21.3s	72%	81	76	76	75	71	76%
67	Cohere Command R+ (Aug. 2024)	$0.020	52.5s	72%	80	77	77	74	72	76%
75	Claude 3.7 Sonnet	$0.040	48.3s	74%	78	77	76	74	74	76%
114	Claude Sonnet 4.6 (Reasoning)	$0.072	1.7m	66%	85	78	74	71	71	76%
110	ByteDance Seed 2.0 Lite	$0.013	2.4m	68%	83	77	74	73	72	76%
74	Qwen 3.5 35B	$0.016	52.3s	71%	80	79	76	74	70	76%
51	Ministral 3 3B	$0.0003	10.1s	70%	80	78	75	72	72	76%
69	Z.AI GLM 4.5	$0.0066	55.8s	71%	79	78	75	74	71	75%
70	Hermes 3 70B	$0.0007	55.2s	70%	79	79	75	73	71	75%
73	WizardLM 2 8x22b	$0.0023	42.7s	69%	82	77	76	70	69	75%
98	Llama 3.1 70B	$0.0023	54.6s	63%	88	74	74	72	67	75%
57	Gemini 2.5 Flash	$0.0065	13.6s	71%	78	77	76	74	69	75%
83	GPT-4o, May 13th (temp=1)	$0.026	13.0s	67%	81	79	76	71	65	74%
84	Z.AI GLM 4.7 Flash	$0.0016	1.0m	68%	82	75	75	70	70	74%
63	GPT-4o, Aug. 6th (temp=0)	$0.018	24.5s	72%	77	75	75	73	72	74%
88	DeepSeek-V2 Chat	$0.0019	47.3s	66%	80	78	73	72	67	74%
92	Stealth: Hunter Alpha	$0.0000	1.2m	68%	78	77	73	73	68	74%
80	LFM2 24B	$0.0002	38.5s	67%	79	77	74	72	66	74%
86	Gemma 3 12B	$0.0005	47.6s	67%	78	78	74	74	64	74%
76	Qwen 3.5 Plus (2026-02-15)	$0.0052	32.0s	69%	79	76	75	72	67	74%
95	Gemini 2.5 Flash (Reasoning)	$0.012	24.8s	65%	82	75	72	70	69	74%
79	Gemini 3 Flash (Preview)	$0.0069	17.7s	66%	81	77	73	70	68	74%
93	GPT-4o, Aug. 6th (temp=1)	$0.018	34.1s	67%	80	73	72	72	71	73%
62	Claude 3 Haiku	$0.0022	15.6s	70%	76	75	73	72	71	73%
66	Gemini 3 Flash (Preview, Reasoning)	$0.010	24.1s	71%	76	75	74	71	71	73%
72	GPT-4o Mini (temp=0)	$0.0012	27.7s	69%	77	74	73	72	71	73%
68	Arcee AI: Trinity Mini	$0.0003	10.5s	68%	77	75	73	73	67	73%
106	Gemini 2.5 Flash Lite (Reasoning)	$0.0032	43.9s	62%	81	79	71	71	64	73%
105	Gemini 3 Pro (Preview)	$0.051	49.5s	70%	75	75	74	72	69	73%
89	Z.AI GLM 4.6	$0.0077	57.0s	69%	75	75	72	72	70	73%
87	Gemma 3 4B	$0.0002	19.2s	66%	76	75	71	70	68	72%
91	Gemini 3.1 Flash Lite (Preview)	$0.0027	7.9s	65%	78	74	72	67	66	71%
94	GPT-4o Mini (temp=1)	$0.0012	31.4s	67%	74	74	71	69	67	71%
113	DeepSeek V3.1	$0.0020	2.4m	67%	75	72	72	69	65	70%
96	GPT-4.1 Mini	$0.0024	18.3s	66%	73	72	70	67	67	70%
97	Inception Mercury	$0.0082	14.5s	67%	74	72	72	66	66	70%
109	GPT-4.1 Nano	$0.0007	12.1s	63%	71	70	67	67	64	68%
107	Inception Mercury 2	$0.0028	7.6s	66%	70	68	68	66	65	67%
108	Stealth: Aurora Alpha	$0.0000	10.2s	65%	70	67	67	66	66	67%
118	Nemotron 3 Super	$0.0000	53.0s	57%	77	69	64	64	61	67%
117	Nemotron 3 Nano	$0.0008	57.0s	59%	72	71	64	63	63	66%
115	GPT-5 Nano	$0.0037	1.2m	63%	70	67	66	65	63	66%
77.61%

Median	Evaluator	Top 3	Flop 3
99.9%	"Not X but Y" pattern overuse	100ByteDance Seed 2.0 Mini 100GPT-4.1 100Gemini 3.1 Pro (Preview)	2GPT-5 Nano 32Nemotron 3 Super 55Stealth: Healer Alpha
55.9%	Adverb-first sentence starts	100Mistral Large 100Mistral Small 4 100GPT-5.4 Mini (Reasoning, Low)	0Inception Mercury 0ByteDance Seed 1.6 0ByteDance Seed 2.0 Lite
80.0%	Adverbs in dialogue tags	100Qwen 3.5 9B 100Cohere Command R+ (Aug. 2024) 100ByteDance Seed 2.0 Mini	1GPT-4.1 Nano 7GPT-4.1 Mini 19Claude 3.7 Sonnet
82.9%	AI-ism adverb frequency	98Inception Mercury 97ByteDance Seed 2.0 Lite 97ByteDance Seed 1.6	52GPT-4.1 Nano 65Claude 3 Haiku 67Gemma 3 4B
100.0%	AI-ism character names	100GPT-4.1 Nano 100GPT-5 Nano 100GPT-4o, May 13th (temp=0)	68Claude Opus 4 76Claude Sonnet 4 80Claude Opus 4.5
100.0%	AI-ism location names	100Llama 3.1 70B 100Ministral 3 3B 100Qwen 3.5 122B	—
44.5%	AI-ism word frequency	85GPT-5.4 82Claude Sonnet 4.6 (Reasoning) 82ByteDance Seed 2.0 Lite	0GPT-4o Mini (temp=0) 0GPT-4o, May 13th (temp=0) 0GPT-4o, Aug. 6th (temp=0)
93.3%	Cliché density	100Qwen 2.5 72B 100MiniMax M2.7 100GPT-5 Mini	40Gemini 2.5 Flash 60Mistral Small 3.2 24B 60Inception Mercury
92.0%	Dialogue tag variety (said vs. fancy)	100Ministral 3B 100GPT-5.4 (Reasoning, Low) 100Mistral Small 3.2 24B	20Claude 3 Haiku 28GPT-4o, Aug. 6th (temp=1) 32Qwen 3.5 Plus (2026-02-15)
24.4%	Em-dash & semicolon overuse	100Qwen 3.5 27B 100Gemini 3.1 Pro (Preview) 100GPT-5.4 Mini (Reasoning, Low)	0DeepSeek V3 (2025-03-24) 0Grok 4 Fast 0Qwen3 235B A22B Instruct 2507
100.0%	Emotion telling (show vs. tell)	100o4 Mini High 100Gemini 3.1 Pro (Preview) 100Mistral Large	73Llama 3.1 8B 86Llama 3.1 70B 90Mistral NeMO
98.4%	Filter word density	100Claude Sonnet 4.5 100GPT-5.4 (Reasoning) 100GPT-4o Mini (temp=1)	11Stealth: Aurora Alpha 22Nemotron 3 Nano 34Inception Mercury 2
100.0%	Gibberish response detection	100GPT-5.4 Mini (Reasoning) 100Gemini 3.1 Flash Lite (Preview) 100Mistral Medium 3.1	46Llama 3.1 8B 52Llama 3.1 70B 77Inception Mercury
100.0%	Markdown formatting overuse	100MiniMax M2.7 100GPT-5 Mini 100Claude 3.5 Sonnet	80Mistral Medium 3.1 98ByteDance Seed 1.6 Flash
100.0%	Missing dialogue indicators (quotation marks)	100Arcee AI: Trinity Mini 100Hermes 3 70B 100Grok 4.20 (Beta)	19Qwen 3.5 397B A17B 40Qwen 3.5 27B 40Gemini 3.1 Pro (Preview)
67.2%	Name drop frequency	100Inception Mercury 2 100GPT-5 Nano 100Arcee AI: Trinity Mini	0GPT-5.1 7Gemma 3 12B 7GPT-5.4 Mini (Reasoning)
68.1%	Narrator intent-glossing	100Cohere Command R+ (Aug. 2024) 100Claude 3 Haiku 99ByteDance Seed 2.0 Lite	0Stealth: Aurora Alpha 0Nemotron 3 Nano 2GPT-5 Nano
100.0%	Overuse of "that" (subordinate clause padding)	100Claude Opus 4 100Qwen 3.5 27B 100Z.AI GLM 4.6	72ByteDance Seed 2.0 Mini 77Claude 3 Haiku 78Claude Haiku 4.5
100.0%	Paragraph length variance	100Claude Opus 4.6 100DeepSeek V3 (2024-12-26) 100Mistral Large	64Inception Mercury 65GPT-5 Nano 66Arcee AI: Trinity Mini
97.0%	Passive voice overuse	100GPT-4o, May 13th (temp=0) 100GPT-4o, Aug. 6th (temp=0) 100o4 Mini High	86WizardLM 2 8x22b 87Arcee AI: Trinity Mini 88Nemotron 3 Super
96.2%	Past progressive (was/were + -ing) overuse	100GPT-4o, May 13th (temp=0) 100Claude 3 Haiku 100Claude Sonnet 4	25Z.AI GLM 4.6 26Claude Sonnet 4.6 40Qwen 3.5 9B
48.2%	Pronoun-first sentence starts	100GPT-5.4 Mini (Reasoning) 99GPT-5.1 95Z.AI GLM 5	0o4 Mini High 0Gemma 3 4B 0DeepSeek V3.1
94.6%	Purple prose (modifier overload)	100Grok 4 100Inception Mercury 100Stealth: Aurora Alpha	76GPT-4.1 Nano 79GPT-5.4 Mini (Reasoning, Low) 82GPT-5.4 Mini
100.0%	Repeated phrase echo	100Stealth: Healer Alpha 100GPT-4o, May 13th (temp=0) 100Z.AI GLM 4.7	—
100.0%	Sentence length variance	100Gemini 2.5 Flash Lite (Reasoning) 100Gemini 3 Pro (Preview) 100GPT-4o Mini (temp=0)	94Inception Mercury 96GPT-4o, Aug. 6th (temp=1) 96Llama 3.1 70B
51.8%	Sentence opener variety	86GPT-4o, Aug. 6th (temp=1) 85Grok 4.1 Fast 85DeepSeek V3 (2025-03-24)	32Qwen 3.5 9B 33Qwen 3.5 397B A17B 34Qwen 3.5 27B
28.5%	Subject-first sentence starts	91Llama 3.1 8B 79Qwen3 235B A22B Instruct 2507 74Mistral Large	0ByteDance Seed 1.6 0Stealth: Aurora Alpha 0Inception Mercury
36.0%	Subordinate conjunction sentence starts	81Mistral Small Creative 80Ministral 3 8B 79Mistral Large	0GPT-4.1 Nano 0Claude 3.5 Sonnet 0LFM2 24B
67.8%	Technical jargon density	100Qwen 2.5 72B 99Mistral Large 99o4 Mini High	0Nemotron 3 Nano 0GPT-5 Nano 4Nemotron 3 Super
45.6%	Useless dialogue additions	100Claude Sonnet 4.6 100Claude Opus 4.6 100GPT-5.4 (Reasoning)	0GPT-4o, May 13th (temp=0) 0Claude 3 Haiku 0Z.AI GLM 4.6

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)