OPEN LINK ↗
// 1h agoRESEARCH PAPER
Claude Opus 5.5 Agents Surface Two Magnetic Candidates
Vals AI reports that 90+ Claude Opus 5.5 agents used DFT simulations to identify YBaMnFeO₅ and KV[Cr(CN)₆] as potential room-temperature, spin-sorting magnetic semiconductors. The findings remain computational: KV[Cr(CN)₆] has been synthesized, but neither candidate’s spin sorting has been experimentally measured.
// ANALYSIS
This is a compelling example of AI accelerating materials research, but it is a validated computational lead—not a laboratory discovery.
- –YBaMnFeO₅ has strong predicted properties, yet its required Mn/Fe checkerboard may scramble during synthesis.
- –KV[Cr(CN)₆] is the more actionable candidate: it was made in 1999 and remained magnetically ordered to 376 K.
- –The public ledger improves credibility by exposing raw Quantum ESPRESSO outputs, reproducibility scripts, and known calculation errors. [GitHub ledger](https://github.com/spicylemonade/compensated-magnet-ledger)
- –The main bottleneck is now experimental validation, not agent-generated hypotheses; community reaction correctly focuses on synthesis, measurement, and novelty claims. [Hacker News discussion](https://news.ycombinator.com/item?id=49970667)
// TAGS
claude-opus-5-5agentllmresearchautomationopen-source
DISCOVERED
1h ago
2026-10-06
PUBLISHED
5h ago
2026-10-05
RELEVANCE
8/ 10
AUTHOR
outlier99