Ritual

Which Ritual Chain feature are you most excited to build with? 🧙

Learn more: ritual.net

Ritual’s mission is to help build the foundations for this next era of collaboration.

Humans created language. With computers, we used code to speak to them.

Yep bro, I know

Come drop by to chat with our interns Rahul, Teo and Arthur, about efficient LLM inference and LLM cognitive inconsistency!

🔮 Ritual is presenting two posters at @NeurIPSConf, San Diego, this Saturday! Spec Dec @ SPIGM 11:30–12:30, Room 20C arxiv.org/abs/2511.15898 LLM Consistency @ Multi-Turn 12:30–1:30, Room 11AB arxiv.org/abs/2511.13240

Preprint: arxiv.org/abs/2511.13240 We had two amazing interns on this project: Teo Kitanovski (github.com/teokitan) and Arthur Liang (@arthlian) Joint work with @ArkaPal999 @akileshpotti @micahgoldblum Follow @ritualnet for more coming soon.

The best LLMs in existence are inconsistent. In our new paper, we measured 7 different open and closed source models – including SOTA such as Gemini 2.5 Pro – and found some shocking results. 🤯