Fantasy: entering an ancient ruin

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)	89%
GPT-5.4 (Reasoning, Low)	86%
GPT-5.4 Mini (Reasoning)	84%
GPT-5.4	84%
GPT-5.4 Mini	83%
GPT-5.4 Mini (Reasoning, Low)	83%
Qwen 3.5 397B A17B	80%
Hermes 3 405B	80%
Gemini 3.1 Pro (Preview)	80%
GPT-5.1	80%
Claude Sonnet 4.6	79%
Qwen 3.5 9B	79%
GPT-4.1	79%
GPT-5	79%
Z.AI GLM 5	79%
Qwen3 235B A22B Instruct 2507	78%
Claude Opus 4.6 (Reasoning)	78%
Rocinante 12B	78%
Mistral Large 2	78%
Grok 4.20 (Beta)	78%

	Score	Cost	Time
GPT-5.4 Mini (Reasoning, Low)	83%	$0.016	18.6s
GPT-5.4 Mini	83%	$0.016	18.8s
Rocinante 12B	78%	$0.0009	26.7s
GPT-5.4 Mini (Reasoning)	84%	$0.024	29.9s
ByteDance Seed 1.6 Flash	76%	$0.0014	29.6s
Mistral Small 4	76%	$0.0015	19.7s
Qwen 3.5 9B	79%	$0.0008	56.9s
GPT-5.4 Nano	77%	$0.0050	17.8s
Qwen3 235B A22B Instruct 2507	78%	$0.0010	56.7s
Z.AI GLM 5	79%	$0.0078	45.3s
Gemini 3.1 Flash Lite (Preview)	76%	$0.0029	8.8s
Hermes 3 405B	80%	$0.0030	1.6m
GPT-4o Mini (temp=1)	77%	$0.0012	28.4s
Claude 3 Haiku	74%	$0.0025	16.9s
o4 Mini	77%	$0.015	23.0s
Grok 4.20 (Beta)	78%	$0.018	19.1s
Z.AI GLM 5 Turbo	77%	$0.0074	34.2s
Mistral Medium 3.1	74%	$0.0048	33.3s
Mistral NeMO	73%	$0.0005	7.9s
DeepSeek-V2 Chat	76%	$0.0023	46.2s

	Score	Consistency	Stability
GPT-5.4 (Reasoning)	89%	95%	85%
GPT-5.4 (Reasoning, Low)	86%	98%	85%
GPT-5.4	84%	96%	81%
GPT-5.4 Mini (Reasoning, Low)	83%	97%	80%
GPT-5.4 Mini	83%	96%	80%
GPT-5.4 Mini (Reasoning)	84%	92%	76%
Qwen 3.5 9B	79%	96%	75%
GPT-5.4 Nano	77%	96%	75%
Grok 4.20 (Beta)	78%	96%	75%
Qwen3 235B A22B Instruct 2507	78%	95%	75%
GPT-5.1	80%	94%	74%
Qwen 3.5 397B A17B	80%	92%	74%
GPT-5.2	77%	97%	74%
o4 Mini	77%	95%	74%
Rocinante 12B	78%	93%	74%
Claude Opus 4.6 (Reasoning)	78%	94%	74%
DeepSeek-V2 Chat	76%	97%	73%
GPT-5 Mini	77%	96%	73%
GPT-5	79%	95%	73%
Grok 4.1 Fast	75%	97%	73%

	Score	Cost	Speed	Stability
GPT-5.4 Mini	83%	$0.016	18.8s	80%
GPT-5.4 Mini (Reasoning, Low)	83%	$0.016	18.6s	80%
GPT-5.4 (Reasoning, Low)	86%	$0.056	1.4m	85%
GPT-5.4 Mini (Reasoning)	84%	$0.024	29.9s	76%
GPT-5.4 (Reasoning)	89%	$0.079	2.2m	85%
GPT-5.4	84%	$0.048	1.4m	81%
Qwen 3.5 9B	79%	$0.0008	56.9s	75%
Rocinante 12B	78%	$0.0009	26.7s	74%
GPT-5.4 Nano	77%	$0.0050	17.8s	75%
Qwen3 235B A22B Instruct 2507	78%	$0.0010	56.7s	75%
Grok 4.20 (Beta)	78%	$0.018	19.1s	75%
Gemini 3.1 Flash Lite (Preview)	76%	$0.0029	8.8s	72%
Hermes 3 405B	80%	$0.0030	1.6m	73%
o4 Mini	77%	$0.015	23.0s	74%
Z.AI GLM 5	79%	$0.0078	45.3s	72%
GPT-5 Mini	77%	$0.010	41.6s	73%
DeepSeek-V2 Chat	76%	$0.0023	46.2s	73%
ByteDance Seed 1.6 Flash	76%	$0.0014	29.6s	72%
Mistral Small 4	76%	$0.0015	19.7s	71%
Mistral Large	76%	$0.014	34.6s	73%

Rank	Model	Avg. Cost	Avg. Time	Stability	# 1	# 2	# 3	# 4	# 5	Total
5	GPT-5.4 (Reasoning)	$0.079	2.2m	85%	93	89	89	89	86	89%
3	GPT-5.4 (Reasoning, Low)	$0.056	1.4m	85%	88	87	86	85	85	86%
4	GPT-5.4 Mini (Reasoning)	$0.024	29.9s	76%	91	87	83	81	80	84%
6	GPT-5.4	$0.048	1.4m	81%	87	86	84	82	81	84%
1	GPT-5.4 Mini	$0.016	18.8s	80%	86	85	83	83	80	83%
2	GPT-5.4 Mini (Reasoning, Low)	$0.016	18.6s	80%	85	84	83	82	80	83%
39	Qwen 3.5 397B A17B	$0.031	2.7m	74%	85	83	81	80	73	80%
13	Hermes 3 405B	$0.0030	1.6m	73%	85	85	80	78	73	80%
68	Gemini 3.1 Pro (Preview)	$0.079	1.6m	73%	87	81	78	78	76	80%
36	GPT-5.1	$0.057	1.4m	74%	83	83	79	78	76	80%
24	Claude Sonnet 4.6	$0.034	47.3s	73%	84	80	77	77	77	79%
7	Qwen 3.5 9B	$0.0008	56.9s	75%	81	81	79	79	75	79%
22	GPT-4.1	$0.018	45.5s	71%	86	77	77	77	76	79%
83	GPT-5	$0.068	2.6m	73%	83	81	78	76	76	79%
15	Z.AI GLM 5	$0.0078	45.3s	72%	83	82	78	78	72	79%
10	Qwen3 235B A22B Instruct 2507	$0.0010	56.7s	75%	82	79	78	77	75	78%
64	Claude Opus 4.6 (Reasoning)	$0.081	1.2m	74%	82	80	78	76	74	78%
8	Rocinante 12B	$0.0009	26.7s	74%	82	81	80	77	72	78%
26	Mistral Large 2	$0.013	31.1s	70%	83	81	76	75	74	78%
11	Grok 4.20 (Beta)	$0.018	19.1s	75%	80	79	78	76	74	78%
16	GPT-5 Mini	$0.010	41.6s	73%	80	78	76	76	75	77%
30	Z.AI GLM 5 Turbo	$0.0074	34.2s	69%	84	81	76	75	69	77%
37	Claude Sonnet 4.5	$0.032	37.0s	70%	83	81	79	75	68	77%
9	GPT-5.4 Nano	$0.0050	17.8s	75%	79	78	78	76	74	77%
28	GPT-4o Mini (temp=1)	$0.0012	28.4s	68%	83	82	76	76	67	77%
89	Claude Opus 4.6	$0.075	1.2m	70%	82	80	77	74	70	77%
56	GPT-5.2	$0.058	1.5m	74%	79	78	77	76	74	77%
14	o4 Mini	$0.015	23.0s	74%	79	79	78	75	73	77%
44	Claude Sonnet 4.6 (Reasoning)	$0.047	1.1m	73%	80	78	77	74	74	76%
79	ByteDance Seed 1.6	$0.013	2.6m	68%	84	78	74	73	73	76%
20	Mistral Large	$0.014	34.6s	73%	80	77	76	76	73	76%
34	Gemini 2.5 Pro	$0.038	38.4s	73%	80	77	76	76	73	76%
18	ByteDance Seed 1.6 Flash	$0.0014	29.6s	72%	80	80	79	72	69	76%
19	Mistral Small 4	$0.0015	19.7s	71%	81	79	78	76	67	76%
17	DeepSeek-V2 Chat	$0.0023	46.2s	73%	79	76	76	75	75	76%
55	Qwen 3.5 35B	$0.0083	34.6s	64%	83	83	72	71	71	76%
12	Gemini 3.1 Flash Lite (Preview)	$0.0029	8.8s	72%	80	77	76	73	73	76%
54	ByteDance Seed 2.0 Lite	$0.012	2.2m	71%	80	79	78	73	67	76%
31	Writer: Palmyra X5	$0.012	23.7s	70%	80	79	76	73	68	75%
23	GPT-5.4 Nano (Reasoning)	$0.0059	21.4s	72%	79	75	75	73	73	75%
57	MiniMax M2.5	$0.0033	1.3m	66%	84	75	73	71	71	75%
32	DeepSeek V3 (2025-03-24)	$0.0015	30.6s	69%	81	77	74	72	72	75%
90	MoonshotAI: Kimi K2.5	$0.018	2.8m	69%	78	78	75	74	68	75%
21	Grok 4.1 Fast	$0.0017	42.4s	73%	76	76	76	74	72	75%
29	Mistral Medium 3.1	$0.0048	33.3s	71%	76	76	76	75	68	74%
25	GPT-5.4 Nano (Reasoning, Low)	$0.0056	19.7s	72%	77	75	74	73	72	74%
58	Qwen 3 32B	$0.0015	56.9s	65%	83	78	73	69	68	74%
117	Claude Opus 4	$0.195	1.4m	69%	78	77	73	73	69	74%
27	Claude 3 Haiku	$0.0025	16.9s	71%	78	77	75	72	69	74%
52	GPT-4o, May 13th (temp=0)	$0.034	16.9s	69%	79	74	73	73	71	74%
59	o4 Mini High	$0.035	59.1s	71%	78	75	75	72	69	74%
33	Claude Haiku 4.5	$0.011	21.1s	70%	77	75	75	73	68	74%
65	DeepSeek V3.2	$0.0014	1.6m	67%	78	76	71	71	71	73%
69	Grok 4.20 (Beta, Reasoning)	$0.038	31.2s	69%	77	76	73	72	69	73%
40	Qwen 3.5 122B	$0.011	29.7s	70%	78	76	75	69	69	73%
35	Ministral 3 14B	$0.0007	18.0s	68%	79	74	73	73	68	73%
41	Stealth: Hunter Alpha	$0.0000	51.0s	69%	78	73	72	72	72	73%
81	Mistral Large 3	$0.0033	32.5s	62%	86	74	71	69	66	73%
80	Hermes 3 70B	$0.0011	1.5m	65%	81	73	72	71	68	73%
71	Llama 3.1 8B	$0.0002	1.2m	65%	81	76	74	72	63	73%
47	Mistral NeMO	$0.0005	7.9s	65%	80	79	74	67	65	73%
45	Gemma 3 12B	$0.0003	51.8s	68%	79	73	73	72	68	73%
43	GPT-4o Mini (temp=0)	$0.0011	28.9s	68%	77	76	72	70	69	73%
46	Mistral Small 4 (Reasoning)	$0.0026	37.5s	68%	78	73	72	71	70	73%
72	Qwen 3.5 Flash	$0.0021	46.0s	64%	81	74	71	70	69	73%
38	Gemini 2.5 Flash Lite	$0.0010	11.1s	68%	78	75	73	70	68	73%
48	GPT-4o, Aug. 6th (temp=1)	$0.017	16.5s	68%	77	75	73	69	69	73%
51	Gemini 2.5 Flash	$0.0061	12.2s	66%	79	73	71	70	70	73%
85	Gemini 3 Pro (Preview)	$0.050	47.5s	70%	74	74	72	72	70	72%
78	Cohere Command R+ (Aug. 2024)	$0.021	51.5s	67%	79	73	72	70	68	72%
49	Qwen 3.5 Plus (2026-02-15)	$0.0059	34.2s	67%	77	75	71	70	70	72%
77	GPT-4o, May 13th (temp=1)	$0.032	16.1s	67%	77	76	73	71	65	72%
88	Claude 3.7 Sonnet	$0.044	53.7s	69%	75	74	73	73	66	72%
112	Claude Opus 4.5	$0.067	53.7s	61%	83	75	70	68	66	72%
110	ByteDance Seed 2.0 Mini	$0.0048	5.3m	70%	74	74	73	70	68	72%
84	Llama 3.1 Nemotron 70B	$0.0031	32.6s	62%	83	71	70	69	66	72%
50	Ministral 8B	$0.0004	11.7s	66%	78	75	71	70	66	72%
53	Z.AI GLM 4.7 Flash	$0.0016	1.2m	69%	73	73	71	71	70	72%
70	Gemma 3 27B	$0.0005	48.8s	65%	79	71	71	71	67	72%
60	Aion 2.0	$0.0064	1.5m	70%	73	72	72	71	70	72%
42	Grok 4 Fast	$0.0018	31.9s	70%	73	73	72	71	68	71%
76	Qwen 3.5 27B	$0.011	1.0m	67%	76	74	73	68	65	71%
66	LFM2 24B	$0.0003	38.2s	66%	76	73	73	71	62	71%
73	GPT-4.1 Mini	$0.0031	19.2s	64%	78	72	71	68	66	71%
94	Z.AI GLM 4.6	$0.0063	1.1m	64%	78	73	70	68	65	71%
75	Qwen 2.5 72B	$0.0010	39.4s	66%	74	73	70	70	64	70%
63	Mistral Small Creative	$0.0008	12.0s	65%	75	73	70	69	66	70%
91	DeepSeek V3 (2024-12-26)	$0.0022	45.8s	63%	77	72	69	68	64	70%
62	Ministral 3 8B	$0.0008	19.6s	66%	74	72	69	69	67	70%
61	Gemini 3 Flash (Preview, Reasoning)	$0.011	25.8s	69%	71	71	70	69	68	70%
93	MiniMax M2.7	$0.0041	1.0m	64%	76	74	72	65	61	70%
92	Gemma 3 4B	$0.0002	16.6s	62%	78	69	67	67	67	70%
86	Gemini 2.5 Flash (Reasoning)	$0.0099	20.4s	65%	75	71	68	68	66	70%
95	DeepSeek V3.1	$0.0022	1.4m	63%	77	73	71	66	61	70%
74	Gemini 3 Flash (Preview)	$0.0065	16.3s	66%	72	72	69	68	66	70%
100	Claude Sonnet 4	$0.030	43.6s	63%	74	74	68	67	64	69%
82	Z.AI GLM 4.5	$0.0061	49.7s	67%	72	69	69	68	68	69%
102	Claude 3.5 Sonnet	$0.047	38.6s	65%	73	73	71	64	64	69%
107	Grok 4	$0.048	1.9m	67%	71	70	70	67	65	69%
67	Ministral 3B	$0.0001	3.5s	66%	71	70	69	67	66	68%
98	Z.AI GLM 4.7	$0.0085	1.1m	63%	74	72	68	65	63	68%
87	Ministral 3 3B	$0.0005	12.1s	64%	73	69	68	65	65	68%
108	GPT-4o, Aug. 6th (temp=0)	$0.017	17.4s	57%	77	71	64	63	63	68%
101	WizardLM 2 8x22b	$0.0021	23.8s	60%	72	70	66	62	60	66%
118	Mistral Small 3.2 24B	$0.012	8.0m	61%	72	68	66	65	60	66%
97	GPT-4.1 Nano	$0.0007	12.3s	62%	70	67	66	65	61	66%
96	Arcee AI: Trinity Mini	$0.0003	10.1s	63%	68	66	65	64	62	65%
103	Gemini 2.5 Flash Lite (Reasoning)	$0.0030	27.6s	61%	69	67	66	62	60	65%
105	Llama 3.1 70B	$0.0010	37.9s	59%	71	66	65	63	58	65%
104	Stealth: Healer Alpha	$0.0000	27.4s	60%	70	68	65	62	60	65%
99	Arcee AI: Trinity Large (Preview)	$0.0000	33.1s	63%	66	66	65	65	61	65%
115	GPT-5 Nano	$0.0042	1.4m	58%	71	65	63	63	59	64%
106	Claude 3.5 Haiku	$0.0033	12.4s	60%	66	66	64	59	58	63%
114	Nemotron 3 Super	$0.0000	1.2m	59%	66	66	62	62	57	63%
113	Nemotron 3 Nano	$0.0009	1.1m	60%	63	63	62	61	59	62%
111	Inception Mercury 2	$0.0030	7.5s	59%	65	63	63	61	56	62%
109	Stealth: Aurora Alpha	$0.0000	9.9s	60%	63	61	61	61	60	61%
116	Inception Mercury	$0.0078	13.9s	53%	63	63	58	56	55	59%
73.09%

Median	Evaluator	Top 3	Flop 3
68.9%	"Not X but Y" pattern overuse	100o4 Mini High 100ByteDance Seed 2.0 Mini 100Gemini 3.1 Pro (Preview)	0Writer: Palmyra X5 0LFM2 24B 0GPT-5 Nano
53.5%	Adverb-first sentence starts	100GPT-5.4 (Reasoning) 100GPT-5.4 Mini (Reasoning, Low) 100GPT-5.4	0Arcee AI: Trinity Mini 0Inception Mercury 2 0Stealth: Aurora Alpha
95.2%	Adverbs in dialogue tags	100Llama 3.1 8B 100o4 Mini 100Nemotron 3 Nano	40Claude Sonnet 4.6 (Reasoning) 52GPT-4.1 Nano 54GPT-5 Nano
84.6%	AI-ism adverb frequency	100ByteDance Seed 1.6 97o4 Mini High 97GPT-5 Mini	55GPT-4.1 Nano 66GPT-4o Mini (temp=0) 70Claude 3.7 Sonnet
100.0%	AI-ism character names	100Claude Sonnet 4.5 100Ministral 8B 100Qwen 3.5 Plus (2026-02-15)	84Llama 3.1 8B 92GPT-4.1 96Gemini 2.5 Flash Lite (Reasoning)
100.0%	AI-ism location names	100Ministral 3B 100GPT-4o, Aug. 6th (temp=0) 100Arcee AI: Trinity Large (Preview)	—
23.5%	AI-ism word frequency	67ByteDance Seed 2.0 Lite 65GPT-5.4 (Reasoning, Low) 64GPT-5	0Gemma 3 4B 0Inception Mercury 2 0GPT-4o, May 13th (temp=1)
100.0%	Cliché density	100Rocinante 12B 100Claude Opus 4 100Claude 3.5 Haiku	60GPT-4o, May 13th (temp=0) 67Stealth: Aurora Alpha 67Inception Mercury
24.5%	Dialogue tag variety (said vs. fancy)	100Claude Sonnet 4.6 100Gemini 3.1 Flash Lite (Preview) 96Claude Sonnet 4.6 (Reasoning)	0GPT-4.1 Nano 0GPT-4o, Aug. 6th (temp=1) 0Gemini 3 Flash (Preview)
1.7%	Em-dash & semicolon overuse	100Mistral Small 3.2 24B 99GPT-4o, May 13th (temp=0) 96GPT-5.4 Mini (Reasoning)	0Grok 4 0Writer: Palmyra X5 0Qwen3 235B A22B Instruct 2507
100.0%	Emotion telling (show vs. tell)	100Gemma 3 12B 100MiniMax M2.7 100GPT-5 Mini	36Llama 3.1 70B 65GPT-4o Mini (temp=0) 68GPT-4o, Aug. 6th (temp=0)
85.3%	Filter word density	100Claude Opus 4.6 100Writer: Palmyra X5 100o4 Mini High	0Inception Mercury 2 0Nemotron 3 Nano 0Stealth: Aurora Alpha
100.0%	Gibberish response detection	100Mistral Small Creative 100Z.AI GLM 4.7 100Arcee AI: Trinity Large (Preview)	40Llama 3.1 8B 80Z.AI GLM 4.6 87Hermes 3 70B
100.0%	Markdown formatting overuse	100Z.AI GLM 5 Turbo 100GPT-5 100GPT-5.4 Mini (Reasoning, Low)	40Mistral Large 3 55Ministral 8B 60Ministral 3 8B
100.0%	Missing dialogue indicators (quotation marks)	100Aion 2.0 100Claude Opus 4 100Claude Sonnet 4	0Gemini 3.1 Flash Lite (Preview) 20Qwen 3.5 397B A17B 60Qwen 3.5 122B
67.5%	Name drop frequency	99GPT-5 Mini 98Stealth: Aurora Alpha 97Gemma 3 4B	10GPT-5.2 20Claude 3 Haiku 27Qwen3 235B A22B Instruct 2507
36.8%	Narrator intent-glossing	100Grok 4.1 Fast 96o4 Mini 92Gemini 3.1 Pro (Preview)	0Stealth: Aurora Alpha 0Arcee AI: Trinity Mini 0Inception Mercury 2
100.0%	Overuse of "that" (subordinate clause padding)	100o4 Mini High 100Mistral Large 100GPT-4o, Aug. 6th (temp=1)	40Mistral Small 3.2 24B 61Llama 3.1 70B 66GPT-4o, Aug. 6th (temp=0)
96.7%	Paragraph length variance	100GPT-5 100Z.AI GLM 5 Turbo 100GPT-5.2	11Mistral Small 3.2 24B 27Inception Mercury 35WizardLM 2 8x22b
97.3%	Passive voice overuse	100Grok 4 Fast 100Grok 4.1 Fast 100DeepSeek V3 (2025-03-24)	83ByteDance Seed 2.0 Mini 83Llama 3.1 70B 86WizardLM 2 8x22b
95.7%	Past progressive (was/were + -ing) overuse	100Grok 4 100GPT-5.1 100Mistral Small 4	14Z.AI GLM 4.7 42Z.AI GLM 4.7 Flash 46Mistral Small Creative
96.5%	Pronoun-first sentence starts	100Z.AI GLM 4.5 100GPT-5.4 Mini (Reasoning) 100Grok 4.20 (Beta)	30Mistral Small 3.2 24B 34Mistral NeMO 64Z.AI GLM 4.7 Flash
96.0%	Purple prose (modifier overload)	100Nemotron 3 Super 100Arcee AI: Trinity Large (Preview) 100Mistral Small 3.2 24B	79Gemini 3.1 Pro (Preview) 80Gemini 2.5 Flash (Reasoning) 84GPT-5.4 (Reasoning, Low)
100.0%	Repeated phrase echo	100Claude Sonnet 4.6 (Reasoning) 100Inception Mercury 100GPT-4o Mini (temp=0)	—
100.0%	Sentence length variance	100Gemini 2.5 Pro 100Z.AI GLM 4.6 100Claude Opus 4	57Mistral Small 3.2 24B 74WizardLM 2 8x22b 82Llama 3.1 70B
55.6%	Sentence opener variety	96Claude 3.5 Haiku 91Rocinante 12B 86Llama 3.1 8B	29Inception Mercury 33GPT-5 Nano 34Mistral Small 3.2 24B
38.3%	Subject-first sentence starts	89GPT-5.4 (Reasoning, Low) 88GPT-5.4 (Reasoning) 86GPT-5.4 Mini (Reasoning, Low)	0Stealth: Aurora Alpha 1Inception Mercury 2Arcee AI: Trinity Mini
40.3%	Subordinate conjunction sentence starts	100Gemini 3.1 Flash Lite (Preview) 92Qwen 3.5 397B A17B 83Gemini 3.1 Pro (Preview)	0Inception Mercury 0Stealth: Aurora Alpha 0Qwen 2.5 72B
34.4%	Technical jargon density	90GPT-5.4 (Reasoning) 86Hermes 3 405B 84Gemma 3 12B	0Grok 4 0Nemotron 3 Nano 0GPT-5 Nano
30.8%	Useless dialogue additions	100GPT-5.4 (Reasoning) 100Gemini 3.1 Flash Lite (Preview) 100Claude Sonnet 4.6	0Z.AI GLM 4.6 0GPT-4.1 Nano 0Arcee AI: Trinity Large (Preview)

Bad Writing Habits

Fantasy: entering an ancient ruin

Performance Score Distribution (Top 20)

Price-Performance Score Distribution (Top 20)

Most Stable Models (Top 20)

Top Overall Models (Top 20)