Last updated: September 12, 2026 at 7:27 a.m. EDT

Writing model ranking based on pairwise results
Rank	Model	AI judge score	Human score
1	Claude Opus 5 💡 adaptive high	93.8	—
2	Kimi K3 💡 enabled	93.6	—
4	GPT-6 Astra 💡 high	92.3	—
5	Claude Fable 5.1 💡 adaptive high	92.1	—
6	GLM 5.3 💡 max	90.5	—
8	Claude Sonnet 5 💡 adaptive high	86.4	—
9	GPT-5.6 Sol 💡 high	85.8	—
10	GLM 5.3 Flash 💡 high	84.0	—
11	MiniMax M3 💡 enabled	81.6	—
12	Muse Spark 1.3 💡 high	80.6	—
14	Hy3 💡 high	79.9	—
15	DeepSeek V4.1 Flash 💡 high	79.3	—
16	Gemini 3.8 Flash 💡 high	78.9	—
19	Qwen3.8 Max 💡 xhigh	73.6	—
22	GPT-5.6 Terra 💡 high	71.2	—
23	Aion 3.0 💡 mandatory	70.1	—
24	Grok 4.6 💡 high	69.2	—
25	Hy4 Preview 💡 high	67.7	—
27	GPT-5.6 Luna 💡 high	65.1	—
29	DeepSeek V4 Pro 💡 enabled	63.1	—
32	Qwen3.8 Flash 💡 high	60.2	—
33	Qwen3.8 27B Q8_0 💡 xhigh** 🏠	59.7	—
38	Gemma 4 31B Q8_0 💡 on* 🏠	56.5	—
40	Inkling 💡 high	55.4	—
42	Qwen3.6 35B A3B Q8_0 💡 on* 🏠	54.2	—
43	MiMo V2.5 Pro 💡 enabled	53.1	—
44	Qwen3.7 Plus 💡 enabled	52.9	—
48	MiMo V2.5 💡 enabled	49.5	—
50	Gemma 4 31B Q8_0 ↯ 🏠	47.1	—
52	Gemma 4 26B A4B Q8_0 💡 on* 🏠	45.8	—
54	Seed 2.1 Turbo ↯	44.5	—
56	Gemma 4 12B Q8_0 💡 on* 🏠	43.8	—
58	Gemma 4 12B Q8_0 ↯ 🏠	39.4	—
60	Granite 4.2 30B Q8_0 💡 full 🏠	34.4	—
62	Gemma 4 26B A4B Q8_0 ↯ 🏠	30.1	—
67	Granite 4.2 8B Q8_0 💡 full 🏠	26.4	—
68	Mistral Medium 3.5 💡 high	25.8	—
71	Qwen3.6 35B A3B Q8_0 ↯ 🏠	24.1	—
72	Nemotron 3 Nano Omni 30B A3B Q8_0 💡 on* 🏠	23.0	—
75	Qwen3.8 27B Q8_0 ↯ 🏠	21.5	—
77	Gemma 4 E2B Q8_0 💡 on* 🏠	20.7	—
81	Gemma 4 E4B Q8_0 ↯ 🏠	15.0	—
83	Gemma 4 E4B Q8_0 💡 on* 🏠	13.3	—
88	Gemma 4 E2B Q8_0 ↯ 🏠	11.4	—

90 models compared52669 published AI verdicts

Method: regularized Bradley–Terry model. The AI-judge score determines rank; the human score is independent and remains secondary until more community votes are available. A score of 50 means an estimated one-in-two chance of beating the average model in this language and category.

Direct duel results

Each cell shows the share of available points earned by the row model against the column model: a win is worth 100%, a tie 50%, and a loss 0%. This result matrix uses only the compatible evaluation protocols included in the main ranking.

💡 marks a model with reasoning, ↯ one without reasoning. 🏠 marks a model run locally on the site author’s computer. “on*” means reasoning is enabled or disabled (local models with no intermediate levels): the requested fine-grained setting is not supported by the local file and falls back to the default enabled mode. For local models with adjustable reasoning (Qwen3.8, Muse Glimmer, Granite 4.2, Nemotron 3 Super), the displayed level (xhigh**, high, full) is the model’s default level: the requested fine-grained setting is not supported by the local file, which falls back to its default behavior.

View as a data table

Points earned in direct duels by model pair
Model	Claude Opus 5 💡 adaptive high	Kimi K3 💡 enabled	GPT-6 Astra 💡 high	Claude Fable 5.1 💡 adaptive high	GLM 5.3 💡 max	Claude Sonnet 5 💡 adaptive high	GPT-5.6 Sol 💡 high	GLM 5.3 Flash 💡 high	MiniMax M3 💡 enabled	Muse Spark 1.3 💡 high	Hy3 💡 high	DeepSeek V4.1 Flash 💡 high	Gemini 3.8 Flash 💡 high	Qwen3.8 Max 💡 xhigh	GPT-5.6 Terra 💡 high	Aion 3.0 💡 mandatory	Grok 4.6 💡 high	Hy4 Preview 💡 high	GPT-5.6 Luna 💡 high	DeepSeek V4 Pro 💡 enabled	Qwen3.8 Flash 💡 high	Qwen3.8 27B Q8_0 💡 xhigh** 🏠	Gemma 4 31B Q8_0 💡 on* 🏠	Inkling 💡 high	Qwen3.6 35B A3B Q8_0 💡 on* 🏠	MiMo V2.5 Pro 💡 enabled	Qwen3.7 Plus 💡 enabled	MiMo V2.5 💡 enabled	Gemma 4 31B Q8_0 ↯ 🏠	Gemma 4 26B A4B Q8_0 💡 on* 🏠	Seed 2.1 Turbo ↯	Gemma 4 12B Q8_0 💡 on* 🏠	Gemma 4 12B Q8_0 ↯ 🏠	Granite 4.2 30B Q8_0 💡 full 🏠	Gemma 4 26B A4B Q8_0 ↯ 🏠	Granite 4.2 8B Q8_0 💡 full 🏠	Mistral Medium 3.5 💡 high	Qwen3.6 35B A3B Q8_0 ↯ 🏠	Nemotron 3 Nano Omni 30B A3B Q8_0 💡 on* 🏠	Qwen3.8 27B Q8_0 ↯ 🏠	Gemma 4 E2B Q8_0 💡 on* 🏠	Gemma 4 E4B Q8_0 ↯ 🏠	Gemma 4 E4B Q8_0 💡 on* 🏠	Gemma 4 E2B Q8_0 ↯ 🏠
Claude Opus 5 💡 adaptive high	—	89.7%	62.5%	68.4%	73.7%	79.3%	86.2%	72.5%	89.7%	84.2%	86.8%	63.6%	92.1%	100.0%	95.0%	81.8%	66.7%	92.9%	95.0%	95.0%	85.7%	86.4%	100.0%	81.8%	95.0%	95.0%	90.0%	95.0%	77.3%	85.0%	95.5%	85.0%	85.0%	85.7%	86.4%	84.6%	100.0%	100.0%	100.0%	100.0%	85.0%	100.0%	100.0%	100.0%
Kimi K3 💡 enabled	10.3%	—	52.5%	34.2%	68.4%	69.0%	86.2%	85.0%	89.7%	89.5%	76.3%	72.7%	81.6%	95.5%	85.0%	90.9%	83.3%	92.9%	80.0%	95.0%	92.9%	95.5%	82.5%	90.9%	92.5%	100.0%	95.0%	100.0%	90.9%	100.0%	100.0%	100.0%	95.0%	100.0%	100.0%	92.3%	100.0%	100.0%	90.9%	100.0%	100.0%	95.0%	100.0%	100.0%
GPT-6 Astra 💡 high	37.5%	47.5%	—	60.0%	35.0%	45.0%	75.0%	42.9%	57.5%	71.4%	72.5%	81.8%	61.9%	72.7%	95.5%	86.4%	86.4%	86.4%	100.0%	95.5%	86.4%	100.0%	100.0%	86.4%	100.0%	95.5%	100.0%	100.0%	95.5%	100.0%	95.5%	100.0%	100.0%	90.9%	100.0%	100.0%	86.4%	100.0%	100.0%	90.9%	100.0%	100.0%	100.0%	100.0%
Claude Fable 5.1 💡 adaptive high	31.6%	65.8%	40.0%	—	76.3%	76.3%	89.5%	60.5%	86.8%	78.9%	84.2%	72.7%	71.1%	95.5%	81.8%	81.8%	81.8%	72.7%	90.9%	90.9%	81.8%	90.9%	90.9%	90.9%	95.5%	86.4%	90.9%	100.0%	92.3%	90.9%	100.0%	100.0%	89.3%	90.9%	96.2%	90.9%	100.0%	92.9%	90.9%	100.0%	100.0%	100.0%	90.9%	100.0%
GLM 5.3 💡 max	26.3%	31.6%	65.0%	23.7%	—	71.1%	73.7%	50.0%	73.7%	81.6%	71.1%	72.7%	63.2%	86.4%	79.2%	90.9%	59.1%	80.8%	87.5%	86.4%	73.1%	77.3%	90.9%	90.9%	86.4%	100.0%	90.9%	100.0%	90.9%	90.9%	100.0%	90.9%	100.0%	100.0%	90.9%	100.0%	100.0%	95.5%	100.0%	100.0%	100.0%	100.0%	100.0%	100.0%
Claude Sonnet 5 💡 adaptive high	20.7%	31.0%	55.0%	23.7%	28.9%	—	63.8%	55.0%	70.7%	68.4%	65.8%	77.3%	78.9%	77.3%	75.0%	90.9%	58.3%	69.2%	80.0%	70.0%	78.6%	77.3%	82.5%	77.3%	90.0%	90.0%	95.0%	90.0%	86.4%	95.0%	90.9%	82.5%	85.0%	85.7%	81.8%	76.9%	100.0%	100.0%	90.9%	100.0%	75.0%	100.0%	100.0%	100.0%
GPT-5.6 Sol 💡 high	13.8%	13.8%	25.0%	10.5%	26.3%	36.2%	—	30.0%	59.0%	52.6%	68.0%	55.0%	68.4%	66.7%	77.5%	63.6%	59.1%	75.0%	80.0%	64.6%	85.7%	86.4%	88.1%	77.3%	91.9%	90.2%	93.8%	97.6%	90.9%	95.5%	81.8%	85.5%	89.5%	96.4%	90.9%	100.0%	100.0%	86.8%	100.0%	100.0%	92.5%	100.0%	100.0%	100.0%
GLM 5.3 Flash 💡 high	27.5%	15.0%	57.1%	39.5%	50.0%	45.0%	70.0%	—	47.5%	57.9%	76.2%	72.7%	76.3%	50.0%	67.9%	65.4%	46.2%	76.9%	75.0%	75.0%	76.9%	85.7%	66.7%	88.5%	73.3%	85.7%	85.7%	92.9%	84.6%	85.7%	100.0%	73.3%	67.9%	84.6%	80.8%	85.7%	92.3%	100.0%	92.3%	100.0%	84.6%	92.9%	93.8%	100.0%
MiniMax M3 💡 enabled	10.3%	10.3%	42.5%	13.2%	26.3%	29.3%	41.0%	52.5%	—	63.2%	60.0%	75.0%	84.2%	79.2%	60.0%	77.3%	66.7%	64.3%	77.5%	61.0%	78.6%	86.4%	81.8%	86.4%	84.8%	75.9%	81.5%	85.2%	90.9%	96.6%	86.4%	83.3%	85.0%	78.6%	100.0%	84.6%	72.7%	100.0%	100.0%	92.3%	80.0%	100.0%	100.0%	100.0%
Muse Spark 1.3 💡 high	15.8%	10.5%	28.6%	21.1%	18.4%	31.6%	47.4%	42.1%	36.8%	—	57.9%	59.1%	55.3%	86.4%	95.5%	68.2%	50.0%	68.2%	86.4%	59.1%	68.2%	81.8%	68.2%	68.2%	81.8%	81.8%	95.5%	90.9%	92.3%	86.4%	72.7%	81.8%	86.7%	90.9%	100.0%	90.9%	86.4%	85.7%	95.5%	100.0%	81.8%	88.5%	90.9%	100.0%
Hy3 💡 high	13.2%	23.7%	27.5%	15.8%	28.9%	34.2%	32.0%	23.8%	40.0%	42.1%	—	36.4%	55.3%	70.8%	72.7%	59.1%	54.2%	76.9%	68.2%	50.0%	69.2%	63.6%	83.3%	86.4%	100.0%	70.8%	79.2%	77.3%	81.8%	93.3%	72.7%	100.0%	90.9%	76.9%	95.5%	96.4%	86.4%	100.0%	100.0%	81.8%	100.0%	90.9%	100.0%	100.0%
DeepSeek V4.1 Flash 💡 high	36.4%	27.3%	18.2%	27.3%	27.3%	22.7%	45.0%	27.3%	25.0%	40.9%	63.6%	—	54.5%	54.5%	59.1%	63.6%	72.7%	63.6%	81.8%	40.0%	59.1%	85.0%	80.0%	59.1%	80.0%	95.5%	86.4%	59.1%	80.0%	85.0%	81.8%	85.0%	85.0%	90.9%	100.0%	86.4%	100.0%	90.0%	100.0%	80.0%	100.0%	95.0%	90.0%	100.0%
Gemini 3.8 Flash 💡 high	7.9%	18.4%	38.1%	28.9%	36.8%	21.1%	31.6%	23.7%	15.8%	44.7%	44.7%	45.5%	—	63.6%	54.5%	63.6%	68.2%	68.2%	63.6%	50.0%	77.3%	73.3%	68.2%	81.8%	72.7%	86.4%	90.9%	81.8%	96.2%	81.8%	81.8%	81.8%	85.7%	100.0%	88.5%	86.4%	95.5%	89.3%	90.9%	100.0%	95.5%	95.5%	100.0%	100.0%
Qwen3.8 Max 💡 xhigh	0.0%	4.5%	27.3%	4.5%	13.6%	22.7%	33.3%	50.0%	20.8%	13.6%	29.2%	45.5%	36.4%	—	59.1%	66.7%	29.2%	46.2%	66.7%	62.5%	50.0%	81.8%	68.2%	75.0%	72.7%	75.0%	75.0%	75.0%	63.6%	81.8%	83.3%	90.9%	77.3%	91.7%	59.1%	87.5%	91.7%	81.8%	100.0%	100.0%	90.9%	90.9%	100.0%	95.5%
GPT-5.6 Terra 💡 high	5.0%	15.0%	4.5%	18.2%	20.8%	25.0%	22.5%	32.1%	40.0%	4.5%	27.3%	40.9%	45.5%	40.9%	—	45.5%	41.7%	71.4%	55.0%	50.0%	50.0%	68.2%	67.5%	54.5%	80.0%	77.5%	92.5%	75.0%	81.8%	85.0%	54.5%	82.5%	80.0%	84.6%	72.7%	78.6%	86.4%	90.0%	95.5%	81.8%	90.0%	95.0%	100.0%	100.0%
Aion 3.0 💡 mandatory	18.2%	9.1%	13.6%	18.2%	9.1%	9.1%	36.4%	34.6%	22.7%	31.8%	40.9%	36.4%	36.4%	33.3%	54.5%	—	41.7%	65.4%	54.5%	50.0%	73.1%	45.5%	58.3%	59.1%	86.4%	81.8%	81.8%	63.6%	72.7%	81.8%	75.0%	77.3%	90.9%	80.8%	72.7%	100.0%	90.9%	95.5%	81.8%	81.8%	100.0%	81.8%	100.0%	100.0%
Grok 4.6 💡 high	33.3%	16.7%	13.6%	18.2%	40.9%	41.7%	40.9%	53.8%	33.3%	50.0%	45.8%	27.3%	31.8%	70.8%	58.3%	58.3%	—	61.5%	66.7%	63.6%	65.4%	68.2%	90.9%	81.8%	72.7%	83.3%	91.7%	66.7%	63.6%	90.9%	75.0%	90.9%	81.8%	70.8%	72.7%	75.0%	100.0%	81.8%	81.8%	63.6%	81.8%	90.9%	95.5%	72.7%
Hy4 Preview 💡 high	7.1%	7.1%	13.6%	27.3%	19.2%	30.8%	25.0%	23.1%	35.7%	31.8%	23.1%	36.4%	31.8%	53.8%	28.6%	34.6%	38.5%	—	35.7%	50.0%	53.8%	64.3%	59.4%	61.5%	46.7%	64.3%	64.3%	57.1%	88.5%	53.1%	46.2%	62.5%	75.0%	96.2%	100.0%	89.3%	100.0%	89.3%	92.3%	84.6%	100.0%	100.0%	92.9%	100.0%
GPT-5.6 Luna 💡 high	5.0%	20.0%	0.0%	9.1%	12.5%	20.0%	20.0%	25.0%	22.5%	13.6%	31.8%	18.2%	36.4%	33.3%	45.0%	45.5%	33.3%	64.3%	—	45.0%	46.2%	35.7%	70.0%	54.5%	57.5%	72.5%	80.0%	57.5%	68.2%	87.5%	54.5%	75.0%	75.0%	92.3%	81.8%	80.8%	72.7%	80.0%	100.0%	83.3%	87.5%	95.0%	100.0%	100.0%
DeepSeek V4 Pro 💡 enabled	5.0%	5.0%	4.5%	9.1%	13.6%	30.0%	35.4%	25.0%	39.0%	40.9%	50.0%	60.0%	50.0%	37.5%	50.0%	50.0%	36.4%	50.0%	55.0%	—	64.3%	63.6%	66.7%	40.9%	63.0%	66.2%	66.2%	75.0%	54.5%	62.5%	90.9%	61.0%	63.4%	75.0%	54.5%	84.6%	91.7%	75.0%	81.8%	81.8%	80.0%	85.0%	95.0%	81.8%
Qwen3.8 Flash 💡 high	14.3%	7.1%	13.6%	18.2%	26.9%	21.4%	14.3%	23.1%	21.4%	31.8%	30.8%	40.9%	22.7%	50.0%	50.0%	26.9%	34.6%	46.2%	53.8%	35.7%	—	50.0%	56.2%	30.8%	53.3%	50.0%	67.9%	39.3%	73.1%	53.6%	69.2%	62.5%	69.2%	92.3%	96.2%	71.4%	84.6%	76.9%	80.8%	76.9%	85.7%	78.1%	78.6%	92.3%
Qwen3.8 27B Q8_0 💡 xhigh** 🏠	13.6%	4.5%	0.0%	9.1%	22.7%	22.7%	13.6%	14.3%	13.6%	18.2%	36.4%	15.0%	26.7%	18.2%	31.8%	54.5%	31.8%	35.7%	64.3%	36.4%	50.0%	—	50.0%	54.5%	55.0%	63.6%	63.6%	45.5%	60.0%	70.0%	72.7%	70.0%	77.3%	75.0%	78.9%	81.2%	100.0%	63.6%	72.7%	72.7%	80.0%	68.2%	100.0%	90.0%
Gemma 4 31B Q8_0 💡 on* 🏠	0.0%	17.5%	0.0%	9.1%	9.1%	17.5%	11.9%	33.3%	18.2%	31.8%	16.7%	20.0%	31.8%	31.8%	32.5%	41.7%	9.1%	40.6%	30.0%	33.3%	43.8%	50.0%	—	53.6%	48.0%	33.9%	42.6%	41.1%	77.3%	68.6%	63.6%	64.0%	77.5%	80.6%	75.0%	100.0%	81.8%	92.5%	89.3%	95.0%	85.0%	100.0%	100.0%	100.0%
Inkling 💡 high	18.2%	9.1%	13.6%	9.1%	9.1%	22.7%	22.7%	11.5%	13.6%	31.8%	13.6%	40.9%	18.2%	25.0%	45.5%	40.9%	18.2%	38.5%	45.5%	59.1%	69.2%	45.5%	46.4%	—	58.3%	70.0%	45.5%	65.0%	68.2%	88.9%	80.0%	77.8%	75.0%	63.6%	65.0%	72.7%	50.0%	90.0%	60.0%	81.8%	83.3%	88.9%	90.0%	90.0%
Qwen3.6 35B A3B Q8_0 💡 on* 🏠	5.0%	7.5%	0.0%	4.5%	13.6%	10.0%	8.1%	26.7%	15.2%	18.2%	0.0%	20.0%	27.3%	27.3%	20.0%	13.6%	27.3%	53.3%	42.5%	37.0%	46.7%	45.0%	52.0%	41.7%	—	63.6%	65.2%	58.7%	60.0%	50.0%	44.8%	60.9%	63.2%	60.7%	90.0%	90.9%	72.6%	87.5%	80.0%	70.0%	82.5%	89.5%	89.5%	100.0%
MiMo V2.5 Pro 💡 enabled	5.0%	0.0%	4.5%	13.6%	0.0%	10.0%	9.8%	14.3%	24.1%	18.2%	29.2%	4.5%	13.6%	25.0%	22.5%	18.2%	16.7%	35.7%	27.5%	33.8%	50.0%	36.4%	66.1%	30.0%	36.4%	—	50.0%	50.0%	72.7%	64.6%	60.0%	78.3%	67.5%	50.0%	88.9%	63.6%	90.0%	75.0%	77.8%	100.0%	75.0%	97.5%	100.0%	100.0%
Qwen3.7 Plus 💡 enabled	10.0%	5.0%	0.0%	9.1%	9.1%	5.0%	6.2%	14.3%	18.5%	4.5%	20.8%	13.6%	9.1%	25.0%	7.5%	18.2%	8.3%	35.7%	20.0%	33.8%	32.1%	36.4%	57.4%	54.5%	34.8%	50.0%	—	58.3%	77.3%	54.2%	36.4%	69.6%	67.5%	66.7%	77.8%	81.8%	63.6%	85.0%	75.0%	63.6%	92.5%	90.0%	95.0%	100.0%
MiMo V2.5 💡 enabled	5.0%	0.0%	0.0%	0.0%	0.0%	10.0%	2.4%	7.1%	14.8%	9.1%	22.7%	40.9%	18.2%	25.0%	25.0%	36.4%	33.3%	42.9%	42.5%	25.0%	60.7%	54.5%	58.9%	35.0%	41.3%	50.0%	41.7%	—	72.7%	52.1%	70.0%	54.3%	80.0%	54.2%	72.2%	50.0%	80.0%	77.5%	83.3%	90.9%	72.5%	90.0%	95.0%	77.8%
Gemma 4 31B Q8_0 ↯ 🏠	22.7%	9.1%	4.5%	7.7%	9.1%	13.6%	9.1%	15.4%	9.1%	7.7%	18.2%	20.0%	3.8%	36.4%	18.2%	27.3%	36.4%	11.5%	31.8%	45.5%	26.9%	40.0%	22.7%	31.8%	40.0%	27.3%	22.7%	27.3%	—	25.0%	45.5%	70.0%	90.0%	73.1%	81.8%	61.5%	72.7%	90.0%	90.9%	81.8%	75.0%	77.3%	90.0%	100.0%
Gemma 4 26B A4B Q8_0 💡 on* 🏠	15.0%	0.0%	0.0%	9.1%	9.1%	5.0%	4.5%	14.3%	3.4%	13.6%	6.7%	15.0%	18.2%	18.2%	15.0%	18.2%	9.1%	46.9%	12.5%	37.5%	46.4%	30.0%	31.4%	11.1%	50.0%	35.4%	45.8%	47.9%	75.0%	—	33.3%	50.0%	57.5%	69.2%	88.9%	76.7%	77.8%	87.5%	83.3%	70.0%	80.0%	95.0%	100.0%	83.3%
Seed 2.1 Turbo ↯	4.5%	0.0%	4.5%	0.0%	0.0%	9.1%	18.2%	0.0%	13.6%	27.3%	27.3%	18.2%	18.2%	16.7%	45.5%	25.0%	25.0%	53.8%	45.5%	9.1%	30.8%	27.3%	36.4%	20.0%	55.2%	40.0%	63.6%	30.0%	54.5%	66.7%	—	61.1%	55.0%	18.2%	80.0%	36.4%	50.0%	55.0%	40.0%	90.9%	88.9%	70.0%	66.7%	90.0%
Gemma 4 12B Q8_0 💡 on* 🏠	15.0%	0.0%	0.0%	0.0%	9.1%	17.5%	14.5%	26.7%	16.7%	18.2%	0.0%	15.0%	18.2%	9.1%	17.5%	22.7%	9.1%	37.5%	25.0%	39.0%	37.5%	30.0%	36.0%	22.2%	39.1%	21.7%	30.4%	45.7%	30.0%	50.0%	38.9%	—	54.2%	57.1%	72.2%	77.3%	88.9%	60.5%	83.3%	69.2%	72.5%	94.7%	94.7%	100.0%
Gemma 4 12B Q8_0 ↯ 🏠	15.0%	5.0%	0.0%	10.7%	0.0%	15.0%	10.5%	32.1%	15.0%	13.3%	9.1%	15.0%	14.3%	22.7%	20.0%	9.1%	18.2%	25.0%	25.0%	36.6%	30.8%	22.7%	22.5%	25.0%	36.8%	32.5%	32.5%	20.0%	10.0%	42.5%	45.0%	45.8%	—	58.3%	75.0%	63.6%	40.0%	58.9%	88.9%	50.0%	80.0%	84.2%	89.5%	77.8%
Granite 4.2 30B Q8_0 💡 full 🏠	14.3%	0.0%	9.1%	9.1%	0.0%	14.3%	3.6%	15.4%	21.4%	9.1%	23.1%	9.1%	0.0%	8.3%	15.4%	19.2%	29.2%	3.8%	7.7%	25.0%	7.7%	25.0%	19.4%	36.4%	39.3%	50.0%	33.3%	45.8%	26.9%	30.8%	81.8%	42.9%	41.7%	—	45.0%	62.5%	60.0%	50.0%	63.6%	73.1%	50.0%	83.3%	57.1%	70.0%
Gemma 4 26B A4B Q8_0 ↯ 🏠	13.6%	0.0%	0.0%	3.8%	9.1%	18.2%	9.1%	19.2%	0.0%	0.0%	4.5%	0.0%	11.5%	40.9%	27.3%	27.3%	27.3%	0.0%	18.2%	45.5%	3.8%	21.1%	25.0%	35.0%	10.0%	11.1%	22.2%	27.8%	18.2%	11.1%	20.0%	27.8%	25.0%	55.0%	—	50.0%	50.0%	55.0%	61.1%	77.3%	80.0%	66.7%	66.7%	90.0%
Granite 4.2 8B Q8_0 💡 full 🏠	15.4%	7.7%	0.0%	9.1%	0.0%	23.1%	0.0%	14.3%	15.4%	9.1%	3.6%	13.6%	13.6%	12.5%	21.4%	0.0%	25.0%	10.7%	19.2%	15.4%	28.6%	18.8%	0.0%	27.3%	9.1%	36.4%	18.2%	50.0%	38.5%	23.3%	63.6%	22.7%	36.4%	37.5%	50.0%	—	40.0%	36.4%	63.6%	46.2%	42.9%	66.7%	63.6%	70.0%
Mistral Medium 3.5 💡 high	0.0%	0.0%	13.6%	0.0%	0.0%	0.0%	0.0%	7.7%	27.3%	13.6%	13.6%	0.0%	4.5%	8.3%	13.6%	9.1%	0.0%	0.0%	27.3%	8.3%	15.4%	0.0%	18.2%	50.0%	27.4%	10.0%	36.4%	20.0%	27.3%	22.2%	50.0%	11.1%	60.0%	40.0%	50.0%	60.0%	—	60.0%	55.0%	50.0%	77.8%	50.0%	55.6%	80.0%
Qwen3.6 35B A3B Q8_0 ↯ 🏠	0.0%	0.0%	0.0%	7.1%	4.5%	0.0%	13.2%	0.0%	0.0%	14.3%	0.0%	10.0%	10.7%	18.2%	10.0%	4.5%	18.2%	10.7%	20.0%	25.0%	23.1%	36.4%	7.5%	10.0%	12.5%	25.0%	15.0%	22.5%	10.0%	12.5%	45.0%	39.5%	41.1%	50.0%	45.0%	63.6%	40.0%	—	60.4%	50.0%	70.0%	76.3%	84.2%	80.0%
Nemotron 3 Nano Omni 30B A3B Q8_0 💡 on* 🏠	0.0%	9.1%	0.0%	9.1%	0.0%	9.1%	0.0%	7.7%	0.0%	4.5%	0.0%	0.0%	9.1%	0.0%	4.5%	18.2%	18.2%	7.7%	0.0%	18.2%	19.2%	27.3%	10.7%	40.0%	20.0%	22.2%	25.0%	16.7%	9.1%	16.7%	60.0%	16.7%	11.1%	36.4%	38.9%	36.4%	45.0%	39.6%	—	63.6%	38.9%	77.8%	66.7%	65.0%
Qwen3.8 27B Q8_0 ↯ 🏠	0.0%	0.0%	9.1%	0.0%	0.0%	0.0%	0.0%	0.0%	7.7%	0.0%	18.2%	20.0%	0.0%	0.0%	18.2%	18.2%	36.4%	15.4%	16.7%	18.2%	23.1%	27.3%	5.0%	18.2%	30.0%	0.0%	36.4%	9.1%	18.2%	30.0%	9.1%	30.8%	50.0%	26.9%	22.7%	53.8%	50.0%	50.0%	36.4%	—	70.0%	25.0%	70.0%	63.6%
Gemma 4 E2B Q8_0 💡 on* 🏠	15.0%	0.0%	0.0%	0.0%	0.0%	25.0%	7.5%	15.4%	20.0%	18.2%	0.0%	0.0%	4.5%	9.1%	10.0%	0.0%	18.2%	0.0%	12.5%	20.0%	14.3%	20.0%	15.0%	16.7%	17.5%	25.0%	7.5%	27.5%	25.0%	20.0%	11.1%	27.5%	20.0%	50.0%	20.0%	57.1%	22.2%	30.0%	61.1%	30.0%	—	50.0%	57.5%	65.0%
Gemma 4 E4B Q8_0 ↯ 🏠	0.0%	5.0%	0.0%	0.0%	0.0%	0.0%	0.0%	7.1%	0.0%	11.5%	9.1%	5.0%	4.5%	9.1%	5.0%	18.2%	9.1%	0.0%	5.0%	15.0%	21.9%	31.8%	0.0%	11.1%	10.5%	2.5%	10.0%	10.0%	22.7%	5.0%	30.0%	5.3%	15.8%	16.7%	33.3%	33.3%	50.0%	23.7%	22.2%	75.0%	50.0%	—	37.5%	50.0%
Gemma 4 E4B Q8_0 💡 on* 🏠	0.0%	0.0%	0.0%	9.1%	0.0%	0.0%	0.0%	6.2%	0.0%	9.1%	0.0%	10.0%	0.0%	0.0%	0.0%	0.0%	4.5%	7.1%	0.0%	5.0%	21.4%	0.0%	0.0%	10.0%	10.5%	0.0%	5.0%	5.0%	10.0%	0.0%	33.3%	5.3%	10.5%	42.9%	33.3%	36.4%	44.4%	15.8%	33.3%	30.0%	42.5%	62.5%	—	40.0%
Gemma 4 E2B Q8_0 ↯ 🏠	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	4.5%	0.0%	0.0%	27.3%	0.0%	0.0%	18.2%	7.7%	10.0%	0.0%	10.0%	0.0%	0.0%	0.0%	22.2%	0.0%	16.7%	10.0%	0.0%	22.2%	30.0%	10.0%	30.0%	20.0%	20.0%	35.0%	36.4%	35.0%	50.0%	60.0%	—

Published duel coverage

Each cell shows how many published AI-judged duels across all published evaluation protocols compare the two models for this language and category. The models on both axes are candidates; this matrix does not show which judge produced each verdict.

View as a data table

Published duel counts by model pair
Model	Claude Opus 5 💡 adaptive high	Kimi K3 💡 enabled	GPT-6 Astra 💡 high	Claude Fable 5.1 💡 adaptive high	GLM 5.3 💡 max	Claude Sonnet 5 💡 adaptive high	GPT-5.6 Sol 💡 high	GLM 5.3 Flash 💡 high	MiniMax M3 💡 enabled	Muse Spark 1.3 💡 high	Hy3 💡 high	DeepSeek V4.1 Flash 💡 high	Gemini 3.8 Flash 💡 high	Qwen3.8 Max 💡 xhigh	GPT-5.6 Terra 💡 high	Aion 3.0 💡 mandatory	Grok 4.6 💡 high	Hy4 Preview 💡 high	GPT-5.6 Luna 💡 high	DeepSeek V4 Pro 💡 enabled	Qwen3.8 Flash 💡 high	Qwen3.8 27B Q8_0 💡 xhigh** 🏠	Gemma 4 31B Q8_0 💡 on* 🏠	Inkling 💡 high	Qwen3.6 35B A3B Q8_0 💡 on* 🏠	MiMo V2.5 Pro 💡 enabled	Qwen3.7 Plus 💡 enabled	MiMo V2.5 💡 enabled	Gemma 4 31B Q8_0 ↯ 🏠	Gemma 4 26B A4B Q8_0 💡 on* 🏠	Seed 2.1 Turbo ↯	Gemma 4 12B Q8_0 💡 on* 🏠	Gemma 4 12B Q8_0 ↯ 🏠	Granite 4.2 30B Q8_0 💡 full 🏠	Gemma 4 26B A4B Q8_0 ↯ 🏠	Granite 4.2 8B Q8_0 💡 full 🏠	Mistral Medium 3.5 💡 high	Qwen3.6 35B A3B Q8_0 ↯ 🏠	Nemotron 3 Nano Omni 30B A3B Q8_0 💡 on* 🏠	Qwen3.8 27B Q8_0 ↯ 🏠	Gemma 4 E2B Q8_0 💡 on* 🏠	Gemma 4 E4B Q8_0 ↯ 🏠	Gemma 4 E4B Q8_0 💡 on* 🏠	Gemma 4 E2B Q8_0 ↯ 🏠
Claude Opus 5 💡 adaptive high	0	29	20	19	19	29	29	20	29	19	19	11	19	11	20	11	12	14	20	20	14	11	20	11	20	20	20	20	11	20	11	20	20	14	11	13	11	20	11	11	20	20	20	11
Kimi K3 💡 enabled	29	0	20	19	19	29	29	20	29	19	19	11	19	11	20	11	12	14	20	20	14	11	20	11	20	20	20	20	11	20	11	20	20	13	11	13	11	20	11	11	20	20	20	11
GPT-6 Astra 💡 high	20	20	0	20	20	20	20	21	20	21	20	11	21	11	11	11	11	11	11	11	11	10	10	11	10	11	11	11	11	11	11	10	11	11	11	11	11	11	11	11	11	10	11	11
Claude Fable 5.1 💡 adaptive high	19	19	20	0	19	19	19	19	19	19	19	11	19	11	11	11	11	11	11	11	11	11	11	11	11	11	11	11	13	11	11	11	14	11	13	11	11	14	11	14	14	13	11	11
GLM 5.3 💡 max	19	19	20	19	0	19	19	21	19	19	19	11	19	11	12	11	11	13	12	11	13	11	11	11	11	12	11	12	11	11	11	11	11	12	11	12	12	11	11	11	11	11	11	11
Claude Sonnet 5 💡 adaptive high	29	29	20	19	19	0	29	20	29	19	19	11	19	11	20	11	12	13	20	20	14	11	20	11	20	20	20	20	11	20	11	20	20	14	11	13	11	20	11	11	20	20	20	11
GPT-5.6 Sol 💡 high	29	29	20	19	19	29	0	20	50	19	25	10	19	12	20	11	11	14	20	41	14	11	42	11	31	41	40	41	11	33	11	31	19	14	11	13	12	19	11	11	20	20	20	11
GLM 5.3 Flash 💡 high	20	20	21	19	21	20	20	0	20	19	21	11	19	13	14	13	13	13	14	14	13	14	15	13	15	14	14	14	13	14	13	15	14	13	13	14	13	14	13	13	13	14	16	18
MiniMax M3 💡 enabled	29	29	20	19	19	29	50	20	0	19	20	10	19	12	20	11	12	14	20	41	14	11	33	11	23	27	27	27	11	29	11	24	20	14	11	13	11	20	11	13	20	20	20	11
Muse Spark 1.3 💡 high	19	19	21	19	19	19	19	19	19	0	19	11	19	11	11	11	11	11	11	11	11	11	11	11	11	11	11	11	13	11	11	11	15	11	14	11	11	14	11	13	11	13	11	12
Hy3 💡 high	19	19	20	19	19	19	25	21	20	19	0	11	19	12	11	11	12	13	11	12	13	11	15	11	10	12	12	11	11	15	11	10	11	13	11	14	11	11	11	11	10	11	10	11
DeepSeek V4.1 Flash 💡 high	11	11	11	11	11	11	10	11	10	11	11	0	11	11	11	11	11	11	11	10	11	10	10	11	10	11	11	11	10	10	11	10	10	11	10	11	11	10	11	10	10	10	10	10
Gemini 3.8 Flash 💡 high	19	19	21	19	19	19	19	19	19	19	19	11	0	11	11	11	11	11	11	11	11	15	11	11	11	11	11	11	13	11	11	11	14	11	13	11	11	14	11	11	11	11	15	14
Qwen3.8 Max 💡 xhigh	11	11	11	11	11	11	12	13	12	11	12	11	11	0	11	12	12	13	12	12	13	11	11	12	11	12	12	12	11	11	12	11	11	12	11	12	12	11	11	13	11	11	11	11
GPT-5.6 Terra 💡 high	20	20	11	11	12	20	20	14	20	11	11	11	11	11	0	11	12	14	20	20	14	11	20	11	20	20	20	20	11	20	11	20	20	13	11	14	11	20	11	11	20	20	20	11
Aion 3.0 💡 mandatory	11	11	11	11	11	11	11	13	11	11	11	11	11	12	11	0	12	13	11	11	13	11	12	11	11	11	11	11	11	11	12	11	11	13	11	12	11	11	11	11	11	11	11	11
Grok 4.6 💡 high	12	12	11	11	11	12	11	13	12	11	12	11	11	12	12	12	0	13	12	11	13	11	11	11	11	12	12	12	11	11	12	11	11	12	11	12	12	11	11	11	11	11	11	11
Hy4 Preview 💡 high	14	14	11	11	13	13	14	13	14	11	13	11	11	13	14	13	13	0	14	14	13	14	16	13	15	14	14	14	13	16	13	16	14	13	12	14	13	14	13	13	13	15	14	12
GPT-5.6 Luna 💡 high	20	20	11	11	12	20	20	14	20	11	11	11	11	12	20	11	12	14	0	20	13	14	20	11	20	20	20	20	11	20	11	20	20	13	11	13	11	20	11	12	20	20	20	11
DeepSeek V4 Pro 💡 enabled	20	20	11	11	11	20	41	14	41	11	12	10	11	12	20	11	11	14	20	0	14	11	27	11	23	40	40	40	11	40	11	41	41	14	11	13	12	20	11	11	20	20	20	11
Qwen3.8 Flash 💡 high	14	14	11	11	13	14	14	13	14	11	13	11	11	13	14	13	13	13	13	14	0	14	16	13	15	14	14	14	13	14	13	16	13	13	13	14	13	13	13	13	14	16	14	13
Qwen3.8 27B Q8_0 💡 xhigh** 🏠	11	11	10	11	11	11	11	14	11	11	11	10	15	11	11	11	11	14	14	11	14	0	10	11	10	11	11	11	10	10	11	10	11	16	19	16	11	11	11	11	10	11	10	10
Gemma 4 31B Q8_0 💡 on* 🏠	20	20	10	11	11	20	42	15	33	11	15	10	11	11	20	12	11	16	20	27	16	10	0	14	25	28	27	28	11	35	11	25	20	18	10	15	11	20	14	10	20	20	20	10
Inkling 💡 high	11	11	11	11	11	11	11	13	11	11	11	11	11	12	11	11	11	13	11	11	13	11	14	0	48	10	11	10	11	9	10	9	10	11	10	11	10	10	10	11	9	9	10	10
Qwen3.6 35B A3B Q8_0 💡 on* 🏠	20	20	10	11	11	20	31	15	23	11	10	10	11	11	20	11	11	15	20	23	15	10	25	48	0	22	23	23	10	20	48	23	19	14	10	11	31	24	10	10	20	19	19	10
MiMo V2.5 Pro 💡 enabled	20	20	11	11	12	20	41	14	27	11	12	11	11	12	20	11	12	14	20	40	14	11	28	10	22	0	31	30	11	24	10	23	20	12	9	11	10	20	9	11	20	20	20	9
Qwen3.7 Plus 💡 enabled	20	20	11	11	11	20	40	14	27	11	12	11	11	12	20	11	12	14	20	40	14	11	27	11	23	31	0	30	11	24	11	23	20	12	9	11	11	20	10	11	20	20	20	9
MiMo V2.5 💡 enabled	20	20	11	11	12	20	41	14	27	11	11	11	11	12	20	11	12	14	20	40	14	11	28	10	23	30	30	0	11	24	10	23	20	12	9	11	10	20	9	11	20	20	20	9
Gemma 4 31B Q8_0 ↯ 🏠	11	11	11	13	11	11	11	13	11	13	11	10	13	11	11	11	11	13	11	11	13	10	11	11	10	11	11	11	0	10	11	10	10	13	11	13	11	10	11	11	10	11	10	10
Gemma 4 26B A4B Q8_0 💡 on* 🏠	20	20	11	11	11	20	33	14	29	11	15	10	11	11	20	11	11	16	20	40	14	10	35	9	20	24	24	24	10	0	9	20	20	13	9	15	9	20	9	10	20	20	20	9
Seed 2.1 Turbo ↯	11	11	11	11	11	11	11	13	11	11	11	11	11	12	11	12	12	13	11	11	13	11	11	10	48	10	11	10	11	9	0	9	10	11	10	11	10	10	10	11	9	10	9	10
Gemma 4 12B Q8_0 💡 on* 🏠	20	20	10	11	11	20	31	15	24	11	10	10	11	11	20	11	11	16	20	41	16	10	25	9	23	23	23	23	10	20	9	0	24	14	9	11	9	19	9	13	20	19	19	9
Gemma 4 12B Q8_0 ↯ 🏠	20	20	11	14	11	20	19	14	20	15	11	10	14	11	20	11	11	14	20	41	13	11	20	10	19	20	20	20	10	20	10	24	0	12	10	11	10	28	9	10	20	19	19	9
Granite 4.2 30B Q8_0 💡 full 🏠	14	13	11	11	12	14	14	13	14	11	13	11	11	12	13	13	12	13	13	14	13	16	18	11	14	12	12	12	13	13	11	14	12	0	10	12	10	12	11	13	14	12	14	10
Gemma 4 26B A4B Q8_0 ↯ 🏠	11	11	11	13	11	11	11	13	11	14	11	10	13	11	11	11	11	12	11	11	13	19	10	10	10	9	9	9	11	9	10	9	10	10	0	10	10	10	9	11	10	9	9	10
Granite 4.2 8B Q8_0 💡 full 🏠	13	13	11	11	12	13	13	14	13	11	14	11	11	12	14	12	12	14	13	13	14	16	15	11	11	11	11	11	13	15	11	11	11	12	10	0	10	11	11	13	14	12	11	10
Mistral Medium 3.5 💡 high	11	11	11	11	12	11	12	13	11	11	11	11	11	12	11	11	12	13	11	12	13	11	11	10	31	10	11	10	11	9	10	9	10	10	10	10	0	10	10	11	9	10	9	10
Qwen3.6 35B A3B Q8_0 ↯ 🏠	20	20	11	14	11	20	19	14	20	14	11	10	14	11	20	11	11	14	20	20	13	11	20	10	24	20	20	20	10	20	10	19	28	12	10	11	10	0	48	10	20	19	19	10
Nemotron 3 Nano Omni 30B A3B Q8_0 💡 on* 🏠	11	11	11	11	11	11	11	13	11	11	11	11	11	11	11	11	11	13	11	11	13	11	14	10	10	9	10	9	11	9	10	9	9	11	9	11	10	48	0	11	9	9	9	10
Qwen3.8 27B Q8_0 ↯ 🏠	11	11	11	14	11	11	11	13	13	13	11	10	11	13	11	11	11	13	12	11	13	11	10	11	10	11	11	11	11	10	11	13	10	13	11	13	11	10	11	0	10	10	10	11
Gemma 4 E2B Q8_0 💡 on* 🏠	20	20	11	14	11	20	20	13	20	11	10	10	11	11	20	11	11	13	20	20	14	10	20	9	20	20	20	20	10	20	9	20	20	14	10	14	9	20	9	10	0	20	20	10
Gemma 4 E4B Q8_0 ↯ 🏠	20	20	10	13	11	20	20	14	20	13	11	10	11	11	20	11	11	15	20	20	16	11	20	9	19	20	20	20	11	20	10	19	19	12	9	12	10	19	9	10	20	0	20	10
Gemma 4 E4B Q8_0 💡 on* 🏠	20	20	11	11	11	20	20	16	20	11	10	10	15	11	20	11	11	14	20	20	14	10	20	10	19	20	20	20	10	20	9	19	19	14	9	11	9	19	9	10	20	20	0	10
Gemma 4 E2B Q8_0 ↯ 🏠	11	11	11	11	11	11	11	18	11	12	11	10	14	11	11	11	11	12	11	11	13	10	10	10	10	9	9	9	10	9	10	9	9	10	10	10	10	10	10	11	10	10	10	0

Help elect the best LLM for writing in English

Which model writes the better text? Judge this anonymous pair and contribute to the human ranking.

Prompt

Write a concise status update (50–150 words) for a community garden committee, summarizing progress on the new toolshed project. Use a short structure with a heading and brief bullet-style points covering: the concrete foundation was poured last weekend, lumber has been ordered and arrives Friday, one volunteer carpenter is still needed for the roof, and the budget remains on track at $1,400 of the $1,600 allotted. End with the target completion date of August 20. Keep the tone factual and organized.

Optional: rate this prompt

1 ★ 2 ★ 3 ★ 4 ★ 5 ★

Response A

Toolshed Project Status

Progress continues on schedule.

- Concrete foundation was poured last weekend and is curing.
- Lumber has been ordered and arrives Friday.
- One volunteer carpenter is still needed for roof installation.
- Budget remains on track at $1,400 of $1,600 allotted.

The committee thanks volunteers for their support.
Target completion date is August 20.

Response B

Toolshed Project Update

Progress to date:
- Foundation: Concrete foundation poured last weekend and is curing on schedule.
- Materials: Lumber has been ordered and is scheduled to arrive Friday.
- Volunteers: One volunteer carpenter is still needed for roof installation.
- Budget: Spending remains on track at $1,400 of the $1,600 allotted.

Target completion date is August 20.

AI judge results

See each AI judge’s detailed ranking and where their results differ.

Muse Glimmer 30B writing test: how it compares to local rivals

Our updated writing test compares both Muse Glimmer quantizations with Gemma, Qwen and Nemotron across four languages, then weighs memory, speed and factual risk.

Model leaderboard