Anthropic says Claude drafted routes to four targets from unpublished syntheses
Anthropic says three Claude models each wrote routes to four targets in 26 to 84 minutes. Professors reviewed the plans blind; none has been tested in a lab.
By RuntimeWire Staff · Published
Primary source: Claude (official)
Why it matters
Route planning consumes expert time before a chemist runs a single reaction, so a model that can draft a credible route from one structure image could change how laboratories allocate graduate-student and postdoc hours. Anthropic's blind design using unpublished targets addresses the recall problem that weakens many chemistry demonstrations.
Anthropic said on November 9th that three Claude models each produced a complete synthetic route to four targets from unpublished syntheses after receiving one image of each target, with no internet, literature or reaction-database access and local code as the only tool. Claude (@claudeai) described the work in a thread on X, and Anthropic published the fuller account in a [report by David…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.