Priori Labs

Nature instilled humans with strong cognitive priors. LLM priors are learned
during pretraining.
On Intelligence LMIQ Benchmark

Research

LMIQ Benchmark

A new benchmark for human level general intelligence capabilities in frontier models.

Ideas

On Intelligence

AGI will be recognized by generality and sample efficient learning ability. ASI will be recognized by capability.

A Sense of Novelty in Large Language Models

Do LLMs have an intrinsic sense of novelty or "interestingness"?

Challenging SOTA Models with a Battleship Game

We tested the ability of frontier AI models to play a novel variant of the classic Battleship game.

Tools

Sokoban Eval Toolkit

An open source Sokoban puzzle builder for evaluating LLM spatial reasoning and multi-step planning.

Dungeon Eval Toolkit

An open source dungeon crawler game for evaluating LLM spatial reasoning and multi-step planning.

Maze Eval Toolkit

An open source maze building toolkit for LLM spatial reasoning evaluations.

Data Exfiltration Sandbox

Sandbox environments for evaluating data exfiltration risks with frontier coding agents.

LLM Think Tank

A multi-stage LLM pipeline tool to plan, review, and synthesize solutions to hard problems.