Dwarkesh spends $20K in API credits making AI models play Blood on the Clocktower; OpenAI beats Anthropic
polynoamial · x · 2026-11-06
Dwarkesh Patel spent $20,000 in API credits having frontier models play each other at Blood on the Clocktower, a social deception game, and turned it into an Elo-ranked benchmark.
Key findings
- Models have improved markedly this year at on-the-spot scheming and bluffing, plus team reasoning to break up conspiracies.
- GPT-6 Astra tops the Elo chart (1860), with GPT-6.1 Sol and GPT-5.6 Sol ahead; Anthropic's Fable 5.1 and Opus 5.5 sit in the 1550–1600 band.
- OpenAI's newer models are clearly better than Anthropic's at this game.
- An OpenAI researcher praised the methodology and half-jokingly offered Dwarkesh an eval-design job.
More from Models
- Google Ultra subscriber blocked from Antigravity by region detection — lxfater · 2026-11-06
- Palisade's Jeff Ladish: skip the 40 benchmarks, ask if agents finish real work — FinanceYF5 · 2026-11-06
- AA Index 31: 594B total / 17B active params, Apache 2.0 open source — AIFlow_ML · 2026-11-06
- Moonshot's Kimi K 3.1 Pro surfaces: 2.8T-param MoE, weights on Hugging Face — thursdai_pod · 2026-11-06
- Tencent Hunyuan Hy4 hits limits on day one, WorkBuddy rushes 10x compute expansion — TencentHunyuan · 2026-11-06
- NaceAI open-sources Drex decision models, hitting SOTA 67.52 on Decision Index — nischay_twt · 2026-11-06