prime-intellect

One command to prepare a rented or local GPU machine for work: rent, connect, clone, pull results, tear down safely.

This Agent Beats AGI? Prime Agent GitHub Tested Locally - Autonomous Recursive Multi-Agent Logic & Puzzle Solver harness inspired by Prime Intellect RLM architecture.

Adversarial QA for LLM-RL environments: find out what reward an empty answer earns. Model-free, zero API cost.

A verifiers RLM environment for testing whether adaptive recursive search outperforms brittle manual RAG choreography on long synthetic corpora.

RL eval environment testing whether a model can pull real action items out of a messy meeting transcript

Automatic titles and persistent color identities for Prime Agent sessions.

A verifiers RL environment that trains models to propose novel, evidence-grounded, falsifiable hypotheses. Rewards novelty with accountability.

Deterministic SRT captioning environment for LLM evaluation and reinforcement learning.

Honesty-shaped finance RL environment with walk-forward DSR and CSCV/PBO grading

RL eval environment testing whether vision models can read analog clocks, gauges, and dials from photos

Single-turn calendar arithmetic RL environment for Prime Intellect: datetime gold solver, exact+format+partial grader, vf-eval ready.

Linear Algebra Done Right (Axler 4e) as an exactly-graded RL environment for the Prime Intellect Environments Hub. 186 tasks, 25 topics, binary reward from exact arithmetic over the rationals. No LLM judge, no network.

RL environment where agents run accounts payable and get graded on the books, not their explanations.

Multi-turn tool-using meeting scheduler RL environment for Prime Intellect: hidden calendars, timezone gold solver, exact+format+partial grader.

Coherence read for Prime Agent (Prime Intellect) sessions. Flags the step where the orchestrator acts on something outside its committed state, such as an undelivered child message.

CanaryBench-90: credential-honeypot distinguishability evaluation for evasion-aware AI coding agents

Prime Intellect RL Environment — Mobile UI Agent with Verifiers-compatible rubric, 88 tests, Docker, LLM eval

Sophos — a Windows-native port of Prime Intellect's Prime Agent. Credits: PrimeIntellect-ai/prime-agent (upstream).