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	Kimi K2.6	1789±105	81% 78%	32
2	Gemini 3.1 Pro Preview	1708±91	86% 56%	36
3	GPT-5.2 (Medium)	1699±62	76% 64%	78
4	GLM 5.1	1690±71	75% 62%	40
5	GPT-5.4 (Low)	1685±57	77% 53%	57
6	Claude Opus 4.6	1629±79	65% 46%	26
7	GPT-5.2 (Low)	1618±58	70% 44%	63
8	Claude Sonnet 4.6 (Low)	1616±56	75% 46%	48
9	Grok 4.1 Fast (Reasoning)	1578±46	71% 38%	85
10	Gemini 3 Flash Preview (Medium)	1565±53	63% 41%	70
11	Kimi K2.5	1548±57	65% 43%	89
12	Gemini 3 Flash Preview (Low)	1507±64	61% 34%	59
13	Qwen 3.5 397B A17B	1493±95	59% 28%	29
14	MiniMax M2.7	1446±79	55% 29%	42
15	Gemini 3.1 Flash-Lite Preview (Low)	1434±59	52% 36%	79
16	Grok 4.1 Fast (Non-reasoning)	1428±53	54% 29%	92
17	GPT-5 mini (Medium)	1413±99	57% 39%	28
18	Claude Haiku 4.5	1406±73	53% 24%	51
19	Gemini 3.1 Flash-Lite Preview (Medium)	1331±67	39% 30%	57
20	Mistral Large 4	1272±94	33% 21%	24
21	GPT-5 mini (Low)	1236±69	24% 24%	46
22	DeepSeek V3.2	1231±94	13% 37%	30
23	Mistral Small 4 (High)	1197±136	16% 19%	31

Good Wins

60%

727 / 1218

Evil Wins

40%

491 / 1218

Slayer Hits

Fake Slayer Shots

203

Monk Blocks

230

Saint Executions

Imp Star Passes

Scarlet Transformations

108

Mayor Wins

Virgin Triggers

109

Ravenkeepers Murdered

127