Launch of Prime Agent: A New Tool for Autonomous AI Sessions
Prime Intellect has launched Prime Agent, a new tool for autonomous AI sessions that significantly improves results on long-context tasks.
Launch of Prime Agent: A New Tool for Autonomous AI Sessions
Prime Intellect has launched Prime Agent — a wrapper that transforms long autonomous AI sessions into a separate programming task.
Its sole tool is a persistent IPython kernel. The model can programmatically search its history, invoke tools, run persistent subagents, and store useful state outside of the active context.
Prime Intellect describes it as follows: "RLM treats context as a variable, and delegating to subagents as function calls within REPL."
At its core are three ideas:
— programmatic invocation of tools via RLM; — persistent operation of multiple subagents; — self-modifying harness that allows the agent to update its own memory, skills, and settings.
They report a noticeable increase in results on tasks with long context and long horizon. On ARC-AGI-3 Opus 5 in standard configuration, it scored 30.2%, while with Prime Agent, the result increased to 95.5%, surpassing the stated human expert baseline of 95.4%.
Also, in a preliminary benchmark, it created working emulators for SEGA Genesis and Game Boy Color from scratch using Rust :D
Why it matters
AnalysisThis tool significantly enhances the efficiency of autonomous AI sessions, which can have a major impact on the development of AI technologies. In particular, it demonstrates the capabilities of creating complex software solutions using AI.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
JSON — Structured Data for Programs, APIs, and AI Agents
A complete hands-on guide to JSON: syntax specifications, serialization via parse and stringify, REST API consumption, JSON Schema validation, and structured outputs for LLMs.
DOCX and XLSX Formats: When AI Needs to Deliver Production-Ready Documents
A comprehensive hands-on guide to generating and processing DOCX documents and XLSX workbooks with AI: comparing against Markdown and CSV, prompt engineering, formula design, and code automation.
Binance publishes investigation into $6.3 billion theft
Binance’s investigation links the theft to a modified Homebrew download, a fake interview and abuse of an internal AI tool.
Reddit user explains why Claude Code weekly limits are burning almost 4x faster
A Reddit user compared token usage across two weeks and found that each 1% of Claude Code's weekly limit now yields far fewer output tokens, pointing to subagents, reasoning time, and UI refresh requests as likely causes.