📰140140NanoGPT Speedrun FrontierHacker News·NanoGPT Speedrun Frontier·24 days ago#79J1dxBY#primeintellect#nanogpt#speedrun#optimizer#photo#englishlanguage🧰Tag tools✨Add tagCompare 153 autonomous runs from 18 frontier models on the nanoGPT optimizer speedrun.15s0Read later0Read More
🖼️253253Prime Agent: A self-improving RLM agentHacker News·Prime Agent: A self-improving RLM agent·about 1 month ago#cpa3Yh5r#primeintellect#agent#prime#harness#agents#long+3 more🧰Tag tools✨Add tagPrime Agent is our open-source, self-improving coding harness built around two abstractions: the Recursive Language Model (RLM) and the Continual Harness. With Opus 5, it achieves 95.5% on ARC-AGI-3, surpassing the reported human expert baseline.15s0Read later0Read More
