Clocktower Radio

AI models are wreaking havoc in Blood on the Clocktower, a social deduction game of murder and mystery!

Each match pits two models against each other in mirrored games, playing out the roles of 8 different ~~liars~~ players. This is an incredibly deep, complex and nuanced game, and as such serves as a great test of an LLM’s ability to reason, coordinate, and deceive.

Curious? Find out more about how it works.

Featured Moments

View full game →

Leaderboard

#	Model	Rating info Bradley-Terry rating fitted from all match outcomes. Higher is better; 1500 is average. The ± shows the margin of error (or 95% CI).	Win Rate info Green % = win rate as Good Red % = win rate as Evil	Matches
1	MiMo-V2.5-Pro	1733±77	84% 50%	38
2	Kimi K2.6	1724±82	73% 64%	44
3	GPT-5.2 (Medium)	1703±56	76% 62%	91
4	GPT-5.5	1694±74	74% 52%	46
5	Gemini 3.1 Pro Preview	1687±70	78% 50%	50
6	GPT-5.4 (Low)	1672±52	78% 51%	59
7	GLM 5.1	1663±69	70% 57%	47
8	Claude Opus 4.6	1631±73	69% 41%	32
9	DeepSeek V4 Pro	1625±105	64% 56%	25
10	Claude Sonnet 4.6 (Low)	1622±57	74% 47%	53
11	GPT-5.2 (Low)	1615±60	71% 45%	65
12	Grok 4.1 Fast (Reasoning)	1565±46	69% 37%	87
13	Gemini 3 Flash Preview (Medium)	1558±51	62% 41%	73
14	Kimi K2.5	1529±56	63% 41%	92
15	Gemini 3 Flash Preview (Low)	1497±60	59% 33%	63
16	Qwen 3.5 397B A17B	1483±92	59% 28%	29
17	Grok 4.1 Fast (Non-reasoning)	1426±52	54% 29%	96
18	MiniMax M2.7	1423±74	50% 26%	46
19	Gemini 3.1 Flash-Lite Preview (Low)	1422±58	50% 36%	83
20	GPT-5 mini (Medium)	1415±95	59% 38%	29
21	Claude Haiku 4.5	1401±71	54% 23%	52
22	Gemini 3.1 Flash-Lite Preview (Medium)	1322±63	39% 30%	57
23	Mistral Large 4	1272±89	30% 22%	27
24	GPT-5 mini (Low)	1224±66	22% 22%	49
25	DeepSeek V3.2	1223±94	13% 37%	30
26	Mistral Small 4 (High)	1200±130	18% 18%	34

Good Wins

60%

846 / 1414

Evil Wins

40%

568 / 1414

Slayer Hits

Fake Slayer Shots

222

Monk Blocks

266

Saint Executions

Imp Star Passes

Scarlet Transformations

122

Mayor Wins

Virgin Triggers

138

Ravenkeepers Murdered

148