Luxas

Luxas is an open-source autonomous research agent that reads papers, runs experiments and writes a compiled LaTeX report from a research question.

Luxas is an open-source multi-agent system for autonomous scientific research. Given a topic, it searches the literature through sources such as OpenAlex, arXiv and CrossRef and reads the papers it finds, then designs and runs experiments in which separate agents write the implementation and the tests without seeing each other's work. It turns raw results into figures, writes a LaTeX report, puts it through adversarial review of content, figures and layout, and produces a compiled PDF with real citations. Runs can last hours and recover from crashes, using file-backed memory and deterministic finish checks. It uses Claude models for most roles and OpenAI o3 for math, can switch to DeepSeek or Kimi, is built on Mario Zechner's pi-mono, and can be tried in the browser at luxas.im.

Luxas at a glance

Best for: Researcher
Runs on: Web, CLI, open source
Pricing: Free and open source; model providers charge for API use
Released: April 2026
Category: AI research tool
Website: luxas.im

Key features

Literature search across OpenAlex, arXiv and CrossRef
Experiments with independent implementation and test agents
Figures generated from raw results
LaTeX report with adversarial review
Compiled PDF with real citations
Crash-recoverable multi-hour runs
Claude, o3, DeepSeek or Kimi models

Use it for

Produce a first-draft research report on a new question
Run and document a small computational study overnight
Survey literature before starting a project

Alternatives to Luxas

OpenResearch — Run parallel research workflows locally or on remote compute.
Arbor — It grows a hypothesis tree and tests changes on held-out data.
Paper2Agent — It enables direct interaction with research papers and automatic result reproduction.
FAROS — Connects literature evidence, executable plans, and human review into an auditable research loop.
Lune — Accesses trusted knowledge from top conference papers for AI agents.
AI Scientist v3 — Automates end-to-end scientific research from task descriptions to complete papers.
ARIS — Cross-model review loops verify research steps without framework lock-in.
Feynman — Read papers, search the web, write research drafts, and cite claims.
Humata — It provides cited answers directly from your uploaded documents.
Open Science — Local-first design with persistent Python and R kernels and traceable artifacts.
LLM Wiki Research — Compiles parallel agent research into source-backed markdown knowledge bases.
Auto Deep Researcher — Automates experiment operations to reduce manual babysitting of training loops.

All Luxas alternatives