PRIME INTELLECT RELEASES PRIME AGENT: AN OPEN-SOURCE RLM HARNESS WHERE SUB-AGENTS ARE FUNCTION CALLS INSIDE PERSISTENT IPYTHON KERNEL

Prime Intellect has released Prime Agent, an open-source coding and research harness that utilizes a Recursive Language Model (RLM) to turn sub-agent calls into functions inside a persistent IPython kernel. According to MarkTechPost, this release is significant as it allows for more efficient and flexible agent development. The Prime Agent has reportedly achieved a 95.5% RHAE Best@1 on ARC-AGI-3, surpassing the human expert baseline.

The Prime Agent harness is built on two key abstractions: the Recursive Language Model and the Continual Harness. The latter enables the agent to edit its own prompts, skills, memory, and sub-agent specs mid-run, allowing for more dynamic and adaptive behavior. MarkTechPost notes that Prime Agent's performance on ARC-AGI-3 exceeds the reported human expert baseline of 95.4%.

The release of Prime Agent is notable in the context of artificial intelligence research, as it provides an open-source tool for developing and testing advanced language models. The ability of Prime Agent to achieve high performance on benchmarks like ARC-AGI-3 suggests its potential for applications in areas like natural language processing and machine learning.

• Prime Intellect releases open-source Prime Agent harness
• Prime Agent utilizes Recursive Language Model and Continual Harness
• Achieves 95.5% RHAE Best@1 on ARC-AGI-3, surpassing human expert baseline

Read the original source for full detail available at MarkTechPost src