alphaXiv launches a preprint server built for AI agents, open-sourcing 12-week archive
SRSchmidgall · x · 2026-11-20
alphaXiv today launched a preprint server built specifically for AI agents. Over three months, nearly 1,000 agents from 35 labs (Stanford, MIT, ETH Zurich and more) conducted research autonomously on it, producing 936 papers alongside code, data, and model weights.
Highlights:
- Agents found significant errors in published human research and created their own journal to review each other's work
- Some independently produced results similar to later human arXiv papers, including failure modes in self-distillation and feature reproducibility in mechanistic interpretability
The team open-sourced the full twelve-week archive of manuscripts, code, data, peer reviews, and chat transcripts.
More from coding & agent
- When agent-built CLIs are too opaque, humans just let the agent run them — craigbalding · 2026-11-20
- Self-hacking with abliterated models: biggest risks were infra side channels, not app code — cyb3rops · 2026-11-20
- Pocket City: a pixel-art Amsterdam canal simulation built entirely on real geodata — anselm · 2026-11-20
- Mnemos vs Claude Code memory test details: 65%/41%/42% recall across three blind tests — RileyRalmuto · 2026-11-20
- Solo-built Mnemos memory system recalls 2x more facts than Claude Code's native memory — RileyRalmuto · 2026-11-20
- Kaggle and Eigent launch AgenticRL-Bench to test coding agents on real RL incidents — dejavucoder · 2026-11-20