Prime Intellect Releases Prime Agent: An Open-Source RLM Harness Where Sub-Agents Are Function Calls Inside Persistent IPython Kernel
Prime Intellect open-sourced an agent framework that treats sub-agents as function calls in a persistent kernel—and hit 95.5% on ARC-AGI-3, above human expert baseline.
Why it matters
A novel agent architecture (RLM + Continual Harness) ships open-source with a credible capability milestone. Practitioners exploring agent design patterns and frameworks have a new reference implementation; the ARC-AGI score signals real reasoning progress.
The key facts
6 to knowPrime Agent: open-source agent harness
Recursive Language Model (RLM) abstraction: sub-agents as functions in persistent IPython kernel
Continual Harness: agent edits own prompts, skills, memory, sub-agent specs mid-run
95.5% RHAE Best@1 on ARC-AGI-3 with Opus 5
Reported human expert baseline: 95.4%
Architecture: persistent stateful execution, self-modification capability
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: Prime Intellect has open-sourced Prime Agent, a coding and research harness built on two abstractions: the Recursive Language Model, which turns sub-agent calls into functions inside a persistent IPython kernel, and the Continual Harness, which lets the agent edit its own prompts, skills, memory,…