The researchers then benchmarked four widely used blastoid-generation protocols, developed by international groups, against that map. They compared molecular identities, developmental timing and lineage structure rather than appearance alone.

How did the models score?

The evaluation revealed substantial differences between methods:

• Some blastoids reproduced all three major cell lineages of a natural human blastocyst relatively well.
• Others misrepresented certain cell types entirely.
• Some models contained large numbers of cells that could not be confidently matched to any known embryonic state.
• No single model perfectly replicated a natural human blastocyst.

"The encouraging finding is that some models capture important aspects of early embryonic development relatively well, although each model has limitations," Yang said.

Why does this matter?

Before this study, the field lacked a consistent yardstick. Looking embryo-like under a microscope says little about whether a model's inner biology matches the real thing.

"Human embryo models have enormous potential for studying the earliest days of an embryo's development, but there has been no consistent way to assess how accurately these reflect real human development," said Yang, who also leads the Trans-Regulatory Biology group at the Charles Perkins Center.

"Our framework allows researchers to compare these models against a detailed biological reference and determine which cell types and developmental processes are faithfully reproduced, and which are not."

The findings come with a clear caution. "Our study also shows that current models are not biologically equivalent to real human embryos, and researchers need to be careful about the conclusions they draw from them," Yang said.

What happens next?

The researchers have made their reference datasets and benchmarking tools publicly available, so scientists anywhere can test new embryo models against the same standards. They hope a shared benchmark will speed up improvements and make evaluations more rigorous across the field.

"If we're going to use these systems to answer important biological questions, we first need to know what they can reliably tell us," Yang said. "Our work provides a roadmap for improving embryo models and ensuring scientific claims remain grounded in what the models can actually support."

The study, by Long, S. et al., appears in Cell Systems (DOI: 10.1016/j.cels.2026.101738).

As with any benchmark, this one reflects the embryo data currently available; the map covers a limited window of early development. But as a first standardized test of its kind, it gives researchers a concrete way to separate what blastoids can faithfully tell us about human beginnings from what they cannot — yet.