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	1728±75	82% 51%	39
2	Kimi K2.6	1720±82	71% 64%	45
3	GPT-5.2 (Medium)	1698±56	76% 60%	93
4	GPT-5.5	1694±74	75% 52%	48
5	Gemini 3.1 Pro Preview	1687±69	79% 50%	52
6	GPT-5.4 (Low)	1678±54	79% 52%	61
7	DeepSeek V4 Pro	1675±88	68% 68%	40
8	GLM 5.1	1665±67	71% 57%	49
9	Claude Opus 4.6	1621±75	67% 39%	33
10	GPT-5.2 (Low)	1620±60	71% 45%	66
11	Claude Sonnet 4.6 (Low)	1614±55	72% 46%	54
12	Grok 4.1 Fast (Reasoning)	1556±46	67% 36%	89
13	Gemini 3 Flash Preview (Medium)	1554±50	60% 41%	75
14	Kimi K2.5	1528±56	63% 41%	93
15	Gemini 3 Flash Preview (Low)	1492±63	58% 33%	64
16	Qwen 3.5 397B A17B	1481±94	59% 28%	29
17	Grok 4.1 Fast (Non-reasoning)	1424±55	54% 29%	98
18	MiniMax M2.7	1420±74	48% 27%	48
19	Gemini 3.1 Flash-Lite Preview (Low)	1418±57	49% 35%	84
20	GPT-5 mini (Medium)	1403±95	55% 35%	31
21	Claude Haiku 4.5	1398±72	54% 23%	52
22	Gemini 3.1 Flash-Lite Preview (Medium)	1320±64	39% 30%	57
23	Mistral Large 4	1276±88	31% 21%	29
24	GPT-5 mini (Low)	1222±63	22% 22%	49
25	DeepSeek V3.2	1221±91	13% 37%	30
26	Mistral Small 4 (High)	1207±129	19% 17%	36

Good Wins

60%

883 / 1478

Evil Wins

40%

595 / 1478

Slayer Hits

Fake Slayer Shots

232

Monk Blocks

273

Saint Executions

Imp Star Passes

Scarlet Transformations

128

Mayor Wins

Virgin Triggers

145

Ravenkeepers Murdered

156