Cognition and Devin factor RSA-260, build GPU siever 10x cheaper than prior art
TheNormanMu · x · 2026-09-10
Cognition researcher Eric Lu, driving a fleet of Devin agents, factored RSA-260 — a new public record for the RSA Factoring Challenge (previous: RSA-250 in Feb 2020).
Key points:
- The Devins built the world's fastest GPU-optimized lattice siever, making factoring 10x cheaper than the previous public state of the art; the project doubled as a proof of concept for Cognition's job scheduler on disaggregated compute.
- Author's estimate: hyperscalers or frontier labs could likely factor RSA-1024 for $30 million per number, possibly substantially less with more optimization; RSA-2048 remains a billion times harder and largely unaffected.
- Context: production RSA keys are 2048-bit (617 digits); 1024-bit RSA was deprecated in 2013.
A landmark case of autonomous agents completing serious research-grade engineering, with real implications for cryptographic cost assumptions.
Related event: Cognition Uses Devin Agent Fleet to Factor RSA-260, Setting New Record(3 posts)→
More from coding & agent
- Resin lets an agent build standalone LLM inference engines; vLLM and SGLang optimize DeepSeek kernels differently — AccBalanced · 2026-10-27
- How I let my cloud Grok Bot code on my Mac mini via Tailscale, no public ports — talkaboutdesign · 2026-10-27
- Context layout is a cost decision: one memory edit can bust your entire agent prompt cache — dscape · 2026-10-27
- Alchemy adds effect/Workflow support across Durable Objects, Workflows and Lambda Durable Functions — samgoodwin89 · 2026-10-27
- Alchemy's Lambda Web Functions run plain Node web servers on Lambda, no OpenNext needed — samgoodwin89 · 2026-10-27
- Alchemy makes all ~2,850 AWS bindings work from any cloud, not just S3/Lambda/DynamoDB — samgoodwin89 · 2026-10-27