We traced the absolute frontier of AI performance and how well can LLMs perform tasks in perfect conditions.

A Lab Building Best Execution for LLMs. Reducing LLM Costs 50% Using Best-Execution for Intelligence.

Python package supporting Llama, Gemma, and Qwen. Control Multiple Behaviors in Language Models at Once

RL-first framework for training coding agents with true online reinforcement learning.

An Unbiased OSS Benchmark For Code Review Agents Measure which tools actually catch bugs and improve code.