Romance: separated couple reunites

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)	90%
GPT-5.4 Mini (Reasoning, Low)	90%
GPT-5.4 Mini (Reasoning)	89%
GPT-5.4 Mini	88%
DeepSeek V3 (2025-03-24)	85%
Claude Opus 4	85%
Rocinante 12B	84%
Z.AI GLM 5 Turbo	84%
GPT-5.1	83%
MiniMax M2.7	83%
Qwen3 235B A22B Instruct 2507	83%
Qwen 3.5 397B A17B	83%
Mistral Large	82%
Mistral Small 4 (Reasoning)	82%
Claude 3.5 Haiku	82%
Qwen 3.5 35B	82%
Mistral Small 4	81%
ByteDance Seed 1.6 Flash	81%

	Score	Cost	Time
Rocinante 12B	84%	$0.0007	17.7s
GPT-5.4 Mini (Reasoning)	89%	$0.019	21.8s
GPT-5.4 Mini (Reasoning, Low)	90%	$0.017	19.1s
GPT-5.4 Mini	88%	$0.018	19.0s
Mistral Small Creative	80%	$0.0007	11.0s
DeepSeek V3 (2025-03-24)	85%	$0.0011	38.6s
Mistral Large	82%	$0.011	28.6s
Mistral Small 4	81%	$0.0016	23.0s
Claude 3.5 Haiku	82%	$0.0021	8.5s
Mistral Small 4 (Reasoning)	82%	$0.0025	37.5s
Writer: Palmyra X5	80%	$0.011	23.9s
ByteDance Seed 1.6 Flash	81%	$0.0012	26.6s
Ministral 3 3B	79%	$0.0002	4.1s
Ministral 3 14B	79%	$0.0005	12.6s
Z.AI GLM 5 Turbo	84%	$0.0082	37.6s
MiniMax M2.7	83%	$0.0039	1.1m
Qwen 3.5 35B	82%	$0.0096	39.3s
Qwen3 235B A22B Instruct 2507	83%	$0.0006	52.1s
Grok 4 Fast	79%	$0.0016	22.9s
Mistral Large 3	81%	$0.0028	32.4s

	Score	Consistency	Stability
GPT-5.4	92%	97%	90%
GPT-5.4 (Reasoning, Low)	90%	97%	88%
GPT-5.4 Mini (Reasoning, Low)	90%	98%	88%
GPT-5.4 (Reasoning)	91%	97%	87%
GPT-5.4 Mini (Reasoning)	89%	96%	86%
GPT-5.4 Mini	88%	96%	85%
DeepSeek V3 (2025-03-24)	85%	96%	81%
Claude Opus 4	85%	96%	80%
Qwen 3.5 397B A17B	83%	96%	79%
Mistral Small 4	81%	97%	79%
Mistral Small 4 (Reasoning)	82%	96%	79%
Qwen 3.5 35B	82%	97%	79%
Qwen3 235B A22B Instruct 2507	83%	93%	78%
MiniMax M2.7	83%	93%	78%
Mistral Medium 3.1	81%	96%	78%
Qwen 3.5 122B	80%	98%	78%
ByteDance Seed 1.6 Flash	81%	96%	77%
GPT-5.1	83%	93%	77%
Qwen 3.5 Flash	80%	97%	77%
Claude Sonnet 4	79%	97%	77%

	Score	Cost	Speed	Stability
GPT-5.4 Mini (Reasoning, Low)	90%	$0.017	19.1s	88%
GPT-5.4 Mini (Reasoning)	89%	$0.019	21.8s	86%
GPT-5.4 Mini	88%	$0.018	19.0s	85%
DeepSeek V3 (2025-03-24)	85%	$0.0011	38.6s	81%
GPT-5.4	92%	$0.065	1.9m	90%
GPT-5.4 (Reasoning, Low)	90%	$0.067	1.7m	88%
Mistral Small 4	81%	$0.0016	23.0s	79%
Mistral Small 4 (Reasoning)	82%	$0.0025	37.5s	79%
ByteDance Seed 1.6 Flash	81%	$0.0012	26.6s	77%
Qwen3 235B A22B Instruct 2507	83%	$0.0006	52.1s	78%
Qwen 3.5 35B	82%	$0.0096	39.3s	79%
Z.AI GLM 5 Turbo	84%	$0.0082	37.6s	76%
Mistral Large	82%	$0.011	28.6s	77%
Rocinante 12B	84%	$0.0007	17.7s	72%
Ministral 3 3B	79%	$0.0002	4.1s	75%
Mistral Small Creative	80%	$0.0007	11.0s	75%
Claude 3.5 Haiku	82%	$0.0021	8.5s	73%
MiniMax M2.7	83%	$0.0039	1.1m	78%
Mistral Large 3	81%	$0.0028	32.4s	76%
Qwen 3.5 Flash	80%	$0.0020	36.1s	77%

Rank	Model	Avg. Cost	Avg. Time	Stability	# 1	# 2	# 3	# 4	# 5	Total
5	GPT-5.4	$0.065	1.9m	90%	94	93	93	91	90	92%
32	GPT-5.4 (Reasoning)	$0.092	2.6m	87%	93	91	90	90	88	91%
6	GPT-5.4 (Reasoning, Low)	$0.067	1.7m	88%	92	91	91	90	88	90%
1	GPT-5.4 Mini (Reasoning, Low)	$0.017	19.1s	88%	91	90	89	89	89	90%
2	GPT-5.4 Mini (Reasoning)	$0.019	21.8s	86%	91	91	90	89	86	89%
3	GPT-5.4 Mini	$0.018	19.0s	85%	91	89	88	86	85	88%
4	DeepSeek V3 (2025-03-24)	$0.0011	38.6s	81%	87	86	84	84	82	85%
110	Claude Opus 4	$0.218	1.5m	80%	88	86	84	82	82	85%
14	Rocinante 12B	$0.0007	17.7s	72%	92	90	88	85	67	84%
12	Z.AI GLM 5 Turbo	$0.0082	37.6s	76%	89	89	83	82	78	84%
47	GPT-5.1	$0.058	1.3m	77%	89	84	83	82	79	83%
18	MiniMax M2.7	$0.0039	1.1m	78%	87	86	84	80	77	83%
10	Qwen3 235B A22B Instruct 2507	$0.0006	52.1s	78%	87	84	84	81	77	83%
38	Qwen 3.5 397B A17B	$0.013	2.1m	79%	86	83	82	82	81	83%
13	Mistral Large	$0.011	28.6s	77%	87	84	83	82	75	82%
8	Mistral Small 4 (Reasoning)	$0.0025	37.5s	79%	85	82	82	81	79	82%
17	Claude 3.5 Haiku	$0.0021	8.5s	73%	89	84	81	78	77	82%
11	Qwen 3.5 35B	$0.0096	39.3s	79%	84	83	82	81	79	82%
7	Mistral Small 4	$0.0016	23.0s	79%	83	82	82	82	78	81%
9	ByteDance Seed 1.6 Flash	$0.0012	26.6s	77%	85	82	80	80	80	81%
36	Claude 3.5 Sonnet	$0.046	36.1s	77%	86	83	82	78	77	81%
63	Claude Opus 4.6	$0.074	1.2m	76%	85	83	81	80	77	81%
19	Mistral Large 3	$0.0028	32.4s	76%	85	82	80	80	77	81%
33	Qwen 3 32B	$0.0014	52.4s	74%	85	84	79	79	76	81%
21	Mistral Medium 3.1	$0.0054	44.5s	78%	83	82	81	80	77	81%
16	Mistral Small Creative	$0.0007	11.0s	75%	85	85	83	76	73	80%
53	Claude Sonnet 4.5	$0.034	43.0s	71%	89	82	80	78	72	80%
30	GPT-4.1	$0.017	35.0s	76%	84	83	80	78	77	80%
27	Writer: Palmyra X5	$0.011	23.9s	75%	84	84	82	76	74	80%
45	Z.AI GLM 5	$0.0083	1.2m	74%	84	83	78	78	77	80%
37	Aion 2.0	$0.0055	1.3m	76%	84	81	80	79	76	80%
107	GPT-5	$0.064	2.5m	72%	85	85	80	79	71	80%
20	Qwen 3.5 Flash	$0.0020	36.1s	77%	82	81	79	79	78	80%
39	Hermes 3 405B	$0.0022	1.6m	77%	82	81	80	79	77	80%
29	Mistral Large 2	$0.012	33.6s	76%	83	82	81	80	74	80%
24	Qwen 3.5 122B	$0.013	38.8s	78%	82	80	79	79	78	80%
64	Claude Opus 4.5	$0.067	53.3s	75%	83	81	80	80	74	80%
34	Ministral 3 14B	$0.0005	12.6s	70%	89	81	80	76	71	79%
15	Ministral 3 3B	$0.0002	4.1s	75%	82	80	79	79	75	79%
23	Ministral 3 8B	$0.0007	21.1s	75%	83	80	78	77	77	79%
52	Claude 3.7 Sonnet	$0.039	45.9s	73%	85	79	78	77	76	79%
22	Grok 4.1 Fast	$0.0016	32.5s	77%	82	80	80	77	76	79%
42	Claude Sonnet 4	$0.034	50.9s	77%	80	80	79	78	77	79%
43	Cohere Command R+ (Aug. 2024)	$0.020	43.0s	74%	81	81	78	78	75	79%
26	Grok 4 Fast	$0.0016	22.9s	75%	82	80	79	79	74	79%
35	Gemma 3 27B	$0.0005	55.2s	75%	82	79	78	78	76	79%
103	Gemini 3.1 Pro (Preview)	$0.076	1.4m	69%	85	84	77	74	73	79%
41	Grok 4.20 (Beta, Reasoning)	$0.030	24.3s	74%	82	79	79	78	74	78%
92	Claude Opus 4.6 (Reasoning)	$0.081	1.3m	75%	82	80	79	75	75	78%
46	Grok 4.20 (Beta)	$0.019	15.8s	71%	83	82	77	75	73	78%
28	GPT-5.4 Nano (Reasoning, Low)	$0.0056	20.6s	75%	81	79	79	78	74	78%
25	GPT-5.4 Nano	$0.0056	20.2s	77%	79	78	78	77	76	78%
71	Arcee AI: Trinity Large (Preview)	$0.0000	58.3s	67%	85	84	76	74	70	78%
101	Grok 4	$0.049	1.8m	70%	83	82	76	75	73	78%
99	DeepSeek V3.2	$0.0013	3.1m	73%	83	80	80	74	72	78%
55	Gemini 2.5 Pro	$0.039	40.6s	74%	82	79	78	76	73	78%
31	GPT-5.4 Nano (Reasoning)	$0.0059	22.2s	75%	80	78	77	76	75	78%
108	Llama 3.1 8B	$0.0005	3.7m	73%	81	78	78	77	72	77%
62	Stealth: Hunter Alpha	$0.0000	1.1m	70%	86	77	76	74	73	77%
40	LFM2 24B	$0.0002	29.8s	72%	82	80	78	75	71	77%
100	GPT-5.2	$0.071	1.8m	75%	79	78	77	76	75	77%
49	o4 Mini High	$0.020	37.2s	73%	80	78	78	77	71	77%
70	Claude Sonnet 4.6 (Reasoning)	$0.037	50.1s	73%	80	77	76	76	74	77%
59	GPT-5 Mini	$0.0097	41.1s	69%	82	80	75	74	73	77%
51	o4 Mini	$0.014	23.5s	70%	82	81	77	75	69	77%
56	Qwen 3.5 27B	$0.011	58.5s	72%	80	78	75	75	74	77%
50	Qwen 3.5 9B	$0.0009	1.2m	73%	79	78	77	75	73	76%
54	MiniMax M2.5	$0.0047	52.2s	71%	81	77	76	74	73	76%
58	Hermes 3 70B	$0.0006	49.8s	70%	81	80	80	74	65	76%
44	Qwen 2.5 72B	$0.0009	37.0s	74%	77	77	75	75	74	76%
72	GPT-4o, May 13th (temp=1)	$0.026	13.7s	68%	81	81	75	71	70	76%
96	ByteDance Seed 2.0 Lite	$0.011	2.2m	72%	78	78	76	73	73	76%
112	MoonshotAI: Kimi K2.5	$0.021	3.1m	70%	82	78	78	70	69	75%
48	Llama 3.1 70B	$0.0008	39.0s	72%	77	77	74	73	73	75%
117	ByteDance Seed 2.0 Mini	$0.0047	5.2m	72%	78	74	74	74	74	75%
75	GPT-4o, May 13th (temp=0)	$0.025	14.4s	68%	83	76	75	72	68	75%
68	Gemini 2.5 Flash	$0.0062	12.8s	67%	80	80	73	72	69	75%
84	DeepSeek V3.1	$0.0022	1.8m	72%	77	76	74	73	73	75%
90	DeepSeek V3 (2024-12-26)	$0.0019	1.0m	66%	83	75	73	72	71	75%
57	Claude Haiku 4.5	$0.011	23.8s	70%	78	78	76	72	68	74%
80	Z.AI GLM 4.7	$0.0092	58.9s	70%	78	74	73	73	72	74%
60	Claude 3 Haiku	$0.0020	15.2s	68%	80	75	73	71	71	74%
66	Gemma 3 12B	$0.0005	34.2s	69%	78	76	75	74	66	74%
65	Stealth: Healer Alpha	$0.0000	25.8s	68%	79	78	76	68	68	74%
82	Z.AI GLM 4.6	$0.0069	46.0s	68%	78	77	75	73	66	74%
67	Ministral 3B	$0.0001	4.9s	66%	82	78	76	67	65	74%
77	Mistral NeMO	$0.0003	9.8s	65%	82	75	73	71	67	74%
73	Gemini 2.5 Flash (Reasoning)	$0.012	25.2s	70%	77	75	73	71	70	73%
83	Claude Sonnet 4.6	$0.030	40.1s	71%	75	75	74	73	69	73%
76	Gemini 2.5 Flash Lite (Reasoning)	$0.0030	37.4s	68%	78	77	75	70	67	73%
113	ByteDance Seed 1.6	$0.016	3.0m	69%	78	76	74	70	69	73%
79	GPT-4o, Aug. 6th (temp=0)	$0.015	23.6s	69%	78	76	74	70	68	73%
94	GPT-4o, Aug. 6th (temp=1)	$0.017	27.8s	66%	79	76	71	70	69	73%
89	Ministral 8B	$0.0004	17.2s	63%	82	75	71	70	67	73%
69	Gemini 3 Flash (Preview)	$0.0072	19.2s	70%	76	74	73	71	69	72%
93	GPT-4o Mini (temp=0)	$0.0012	32.7s	65%	81	74	71	68	68	72%
74	Arcee AI: Trinity Mini	$0.0003	10.1s	67%	78	75	75	69	64	72%
106	Gemini 3 Pro (Preview)	$0.053	53.4s	69%	75	74	71	71	70	72%
98	Z.AI GLM 4.5	$0.0050	45.5s	66%	78	74	71	70	67	72%
61	Gemini 2.5 Flash Lite	$0.0009	12.4s	70%	74	73	72	71	70	72%
88	Qwen 3.5 Plus (2026-02-15)	$0.0052	29.5s	66%	78	74	72	70	66	72%
85	DeepSeek-V2 Chat	$0.0018	50.7s	68%	75	73	73	73	66	72%
81	Llama 3.1 Nemotron 70B	$0.0023	35.4s	70%	73	72	72	71	68	71%
91	Mistral Small 3.2 24B	$0.0006	19.3s	65%	78	76	76	68	59	71%
87	Gemini 3 Flash (Preview, Reasoning)	$0.011	26.9s	68%	74	73	72	71	65	71%
78	GPT-4.1 Mini	$0.0023	14.4s	68%	73	72	71	70	69	71%
95	GPT-4o Mini (temp=1)	$0.0012	29.9s	66%	74	73	69	69	68	71%
86	Gemini 3.1 Flash Lite (Preview)	$0.0027	8.7s	65%	76	73	72	67	64	71%
109	Z.AI GLM 4.7 Flash	$0.0018	1.5m	65%	73	71	69	67	67	69%
97	Gemma 3 4B	$0.0002	16.7s	65%	72	72	70	70	63	69%
104	GPT-4.1 Nano	$0.0006	11.1s	63%	70	69	66	66	64	67%
114	GPT-5 Nano	$0.0039	1.2m	62%	72	68	66	65	64	67%
102	WizardLM 2 8x22b	$0.0016	24.8s	65%	69	68	67	66	65	67%
105	Stealth: Aurora Alpha	$0.0000	10.0s	62%	72	69	67	64	63	67%
111	Inception Mercury	$0.0061	12.3s	59%	74	71	66	62	61	67%
115	Nemotron 3 Super	$0.0000	1.2m	61%	69	69	66	63	60	66%
116	Inception Mercury 2	$0.0029	7.6s	59%	65	62	62	61	58	62%
118	Nemotron 3 Nano	$0.0009	1.2m	58%	64	62	60	60	58	61%
76.88%

Median	Evaluator	Top 3	Flop 3
100.0%	"Not X but Y" pattern overuse	100GPT-4o Mini (temp=0) 100Inception Mercury 100Grok 4	32GPT-5 Nano 51Nemotron 3 Nano 53Stealth: Healer Alpha
55.3%	Adverb-first sentence starts	100GPT-5.4 Mini (Reasoning) 100GPT-5.4 Mini (Reasoning, Low) 100GPT-5.4 Mini	2Inception Mercury 6Inception Mercury 2 8Nemotron 3 Nano
90.9%	Adverbs in dialogue tags	100GPT-5.4 100Qwen 3.5 27B 100Grok 4.1 Fast	0GPT-4.1 Nano 32GPT-5 Nano 35Claude Haiku 4.5
82.6%	AI-ism adverb frequency	99ByteDance Seed 2.0 Lite 97ByteDance Seed 1.6 96Grok 4.1 Fast	51Claude 3 Haiku 53Mistral Small 3.2 24B 53GPT-4.1 Nano
100.0%	AI-ism character names	100Gemini 3.1 Pro (Preview) 100ByteDance Seed 1.6 Flash 100DeepSeek V3 (2024-12-26)	64Llama 3.1 8B 92Claude Sonnet 4 92Grok 4.20 (Beta)
100.0%	AI-ism location names	100GPT-5.4 Nano (Reasoning, Low) 100Stealth: Aurora Alpha 100GPT-4o Mini (temp=0)	96DeepSeek V3.2 96o4 Mini
47.2%	AI-ism word frequency	85Claude Sonnet 4.6 84ByteDance Seed 1.6 Flash 83ByteDance Seed 2.0 Lite	0GPT-4o Mini (temp=1) 0Llama 3.1 Nemotron 70B 0GPT-4o Mini (temp=0)
93.3%	Cliché density	100Claude Sonnet 4.6 100Gemini 3 Flash (Preview) 100Grok 4.1 Fast	33Qwen 2.5 72B 47Inception Mercury 2 53Mistral NeMO
72.8%	Dialogue tag variety (said vs. fancy)	100Claude Sonnet 4.6 (Reasoning) 100GPT-5.4 (Reasoning) 100Mistral Small Creative	4Gemini 3 Flash (Preview, Reasoning) 4Gemini 2.5 Flash (Reasoning) 6Grok 4 Fast
20.0%	Em-dash & semicolon overuse	100Qwen 3.5 397B A17B 100GPT-5.4 Mini 100Qwen 3.5 122B	0GPT-4.1 Nano 0o4 Mini High 0GPT-4.1
100.0%	Emotion telling (show vs. tell)	100Ministral 8B 100Gemini 3 Flash (Preview, Reasoning) 100Claude Sonnet 4.5	83Llama 3.1 70B 83Inception Mercury 89GPT-4o, May 13th (temp=0)
94.5%	Filter word density	100Claude Opus 4.6 (Reasoning) 100GPT-4o, May 13th (temp=1) 100GPT-4.1 Mini	2Nemotron 3 Nano 9Inception Mercury 2 16Stealth: Aurora Alpha
100.0%	Gibberish response detection	100GPT-5.4 (Reasoning) 100Gemini 3.1 Flash Lite (Preview) 100DeepSeek V3.1	0Llama 3.1 8B 80Inception Mercury 80Llama 3.1 70B
100.0%	Markdown formatting overuse	100Stealth: Healer Alpha 100Ministral 3 8B 100Mistral Small 3.2 24B	80Ministral 3B
100.0%	Missing dialogue indicators (quotation marks)	100Grok 4 Fast 100Grok 4.20 (Beta) 100Stealth: Healer Alpha	22Qwen 3.5 35B 26Qwen 3.5 397B A17B 60Gemini 3.1 Pro (Preview)
89.7%	Name drop frequency	100Nemotron 3 Super 100GPT-4.1 Nano 100Stealth: Hunter Alpha	21Mistral Large 29Mistral Large 2 30GPT-5.4 Nano
71.5%	Narrator intent-glossing	100GPT-4o Mini (temp=0) 100Gemini 3 Pro (Preview) 100Gemini 3.1 Pro (Preview)	0Inception Mercury 2 0Nemotron 3 Nano 0GPT-5 Nano
100.0%	Overuse of "that" (subordinate clause padding)	100Gemini 3.1 Flash Lite (Preview) 100Grok 4 100Gemini 3 Pro (Preview)	42Claude 3.5 Haiku 49Mistral Small 3.2 24B 50ByteDance Seed 2.0 Mini
100.0%	Paragraph length variance	100DeepSeek V3.2 100Claude Opus 4.5 100GPT-5.4 (Reasoning)	47WizardLM 2 8x22b 51GPT-5 Nano 59Nemotron 3 Nano
98.2%	Passive voice overuse	100Gemma 3 12B 100Claude Opus 4.5 100Grok 4 Fast	83Inception Mercury 91ByteDance Seed 2.0 Mini 92ByteDance Seed 2.0 Lite
95.0%	Past progressive (was/were + -ing) overuse	100Llama 3.1 8B 100GPT-5 100ByteDance Seed 1.6 Flash	31MiniMax M2.5 33Claude Sonnet 4.6 37Ministral 3B
29.4%	Pronoun-first sentence starts	100Llama 3.1 8B 100Llama 3.1 Nemotron 70B 98Mistral Large	0o4 Mini High 0Gemini 2.5 Flash (Reasoning) 0Z.AI GLM 4.7 Flash
94.3%	Purple prose (modifier overload)	100Nemotron 3 Nano 100Stealth: Aurora Alpha 100Cohere Command R+ (Aug. 2024)	73Llama 3.1 8B 79GPT-4.1 Nano 79Gemma 3 4B
100.0%	Repeated phrase echo	100o4 Mini 100Llama 3.1 8B 100Qwen 3 32B	—
100.0%	Sentence length variance	100Grok 4.1 Fast 100DeepSeek V3 (2024-12-26) 100Qwen 3.5 397B A17B	91WizardLM 2 8x22b 96GPT-4o, Aug. 6th (temp=1) 97Inception Mercury
50.7%	Sentence opener variety	97Claude 3.5 Haiku 95Llama 3.1 8B 89Rocinante 12B	30Inception Mercury 33Qwen 3.5 35B 34Qwen 3.5 9B
29.3%	Subject-first sentence starts	91Rocinante 12B 87GPT-5.4 80Llama 3.1 8B	2Qwen 3.5 Plus (2026-02-15) 2Inception Mercury 2 3GPT-5 Nano
32.2%	Subordinate conjunction sentence starts	79Llama 3.1 8B 74GPT-5.4 Mini (Reasoning, Low) 73GPT-5.4 (Reasoning, Low)	0Mistral Large 2 0MiniMax M2.5 0Claude Sonnet 4.6
66.5%	Technical jargon density	100GPT-5.4 (Reasoning) 100DeepSeek V3 (2025-03-24) 96GPT-5.1	0Inception Mercury 2 0Nemotron 3 Super 0GPT-5 Nano
31.6%	Useless dialogue additions	99Qwen 3.5 35B 99Qwen 3.5 397B A17B 99GPT-5.4 Mini (Reasoning, Low)	0DeepSeek-V2 Chat 0Stealth: Aurora Alpha 0Inception Mercury 2

Bad Writing Habits

Romance: separated couple reunites

Performance Score Distribution (Top 20)

Price-Performance Score Distribution (Top 20)

Most Stable Models (Top 20)

Top Overall Models (Top 20)