Microsoft open-sources RetroChimera for AI molecule synthesis

Microsoft Research published RetroChimera — an AI retrosynthesis model in Nature that proposes lab synthesis routes for target molecules — and open-sourced the implementation and weights under MIT, with validation that includes transfer to proprietary chemistry data from collaborators including GSK and Novartis.

HEALTH desk — designing a molecule is only half the problem; AI that plans how to make it in the lab is the bottleneck drug and materials teams actually hit.

The job is planning a recipe, not inventing a medicine. Microsoft says designing a new molecule is no longer the only hard part. Making it in the lab is still slow, manual, and expensive. RetroChimera proposes a route: a sequence of reactions from purchasable building blocks up to the target. A building block is a chemical you can order. A reaction is one lab step that turns one set of chemicals into the next. The blog’s Figure 1 is that backward path. File that job as Microsoft’s. A proposed route is not a drug, and it is not a batch this desk watched a chemist run.

The model is an ensemble of two parts that fail in different ways, then a learned vote. An ensemble means more than one model, and a rule for combining them. R-SMILES 2 is a Transformer — a text-style model — that writes the precursor molecules straight from the target. Microsoft says that freedom lets it learn reaction patterns from data, and that the same freedom can make it invent a step that is not real. NeuralLoc is a graph model. It treats the molecule as a connected drawing, not a line of text. It picks a reaction template, a stored pattern of a known reaction, and predicts where on the target that pattern should land. Microsoft says NeuralLoc stays closer to patterns in the training data, and is weaker when the needed reaction is not in that library. R-SMILES 2 does better when the molecule changes a lot in one step. NeuralLoc does better on rarer reactions and on smaller, local changes. File those roles as the blog’s.

The vote is the third piece. Each model gives a rank-dependent vote to each set of reactants it proposes. A reactant is an input chemical for one step. If both models propose the same reaction, the votes add. Microsoft says the system learns how much to trust each model at each rank, so it can stay close to whichever sub-model is better for that kind of reaction. The blog’s Figure 2 is that picture. The blog also says the same frame could later take other sources, such as a reaction database or a chemist in the loop. That extra source is drawn as a possibility. It is not what this release says is running. File the vote as the blog’s.

Who else is on the paper. Microsoft Source says the work is from Microsoft Research and collaborators including GSK and Novartis, two large drug companies. The Nature article is titled “Chemist-aligned retrosynthesis by ensembling diverse inductive bias models,” published 21 Sep 2026, DOI 10.1038/s41586-026-11160-9. The abstract says the authors show zero-shot transfer and fine-tuning on internal datasets from two major pharmaceutical companies. Zero-shot, here, means trying the already-trained model on a company’s private chemistry without a new training pass first. Fine-tuning means a further training pass on that private set. The abstract does not name the two companies in that sentence. The author list on the same page includes people at GSK in Stevenage and at Novartis Biomedical Research in Basel, plus Microsoft Research. Source says the pre-trained model can be readily adapted to GSK’s internal data and its proprietary chemistry data. File the two-company transfer line as the abstract’s. File the GSK and Novartis names as Source’s, and as the affiliations printed on the Nature page. This desk did not see the private datasets.

The chemist check on full routes, still Microsoft. Source says expert chemists looked at proposed pathways for 10 molecules chosen to compare RetroChimera with other models. RetroChimera produced a fully accepted sequence for nine. The other models landed between two and five. The blog’s Figure 3 is more specific: nine targets for RetroChimera, five for the de novo model, four for the editing model, and two for NeuralSym, which the blog calls a strong baseline. Ten molecules is a review set. It is not a catalog of every drug target. File the nine-of-ten line and the two-to-five range as Source’s. File the named five, four, and two as the blog’s figure summary. This desk did not score the routes.

A second chemist check is a blind preference. Source says nine Ph.D.-level organic chemists, from Microsoft and from major pharmaceutical companies, were given a choice between RetroChimera’s top suggestion and a previously documented way of making the same molecule. They preferred RetroChimera about 64% of the time. About 64% means a bit under two in three, on that comparison, as Source reports it. The Nature abstract says organic chemists preferred RetroChimera over published reference reactions and over other AI models, in both pairwise and pointwise setups. Pairwise means a head-to-head choice. Pointwise means scoring one suggestion on its own. The 64% figure is Source’s. The abstract does not print 64. File each number with the page that prints it.

Where the weights live. The blog says RetroChimera is on GitHub under the MIT license, and also accessible via Microsoft Foundry. The MIT license is a short permission that lets other people use, change, and share the code, including in commercial work, as long as they keep the license notice. The repository LICENSE file is the MIT License, copyright Microsoft Corporation. The README’s license badge says MIT. Prefer that GitHub release as the shipping fact. Foundry is a second door the blog names. This filing does not treat a Foundry login as the open-source release. The README says the model can still be wrong, that lower-ranked suggestions are more likely to be made up, and that a chemist has to check any prediction before it is used in a real lab. It also says the main checkpoint was trained on reaction data available through 2023. File those limits as the README’s. Install commands stay on that page.

Do not read this as a new drug. Do not read it as FDA approval, a clinical trial, or a named molecule this desk is claiming was made. Microsoft and Nature do not say RetroChimera invented a medicine. A route a model prefers is not a pill on a shelf. Do not treat the ten-target review as a success rate for every chemistry problem.

Plain English for the rest of the card: retrosynthesis = start from the molecule you want and work backward to pieces you can buy. ensemble = two models plus a rule that combines them. Transformer = a model that writes the next chemicals as text. graph model = a model that reads the molecule as a connected drawing. template = a stored pattern of a known reaction. zero-shot = try the trained model on new private data without retraining first. fine-tuning = a further training pass on that private data. MIT license = permission to use and share the code, including commercially, with the license notice kept. building block = a chemical you can order. This filing is the 21 Sep Nature paper plus the MIT weights. It is not a drug approval.

PRIMARY here: the Microsoft Research blog of 21 Sep 2026, “Improving synthesis prediction of small molecules at scale with RetroChimera” — Tier A PRIMARY company source, the originating research post — the Microsoft Source feature the same day, the Nature article published 21 Sep 2026, and the GitHub repository under the MIT license. The Nature publication, the open-source implementation and weights, the backward-planning job, the R-SMILES 2 and NeuralLoc split, the learned vote, and the Figure 3 counts of nine, five, four, and two are the blog’s. The GSK and Novartis collaborator names, the GSK internal-data line, the nine-of-ten fully accepted routes, the two-to-five range, and the about-64% blind preference are Source’s. The two-company zero-shot and fine-tuning line, the paper title, and the GSK and Novartis author affiliations are the Nature page’s. The MIT license text and the chemist-must-check warning are the repository’s. NOT claimed: that RetroChimera invents a new drug, FDA approval, a clinical outcome, a molecule name, that this desk ran the model or saw the private datasets, that Foundry is the open-source release, a stock tip, or investment advice. Distinct from the already-filed fda-nams-nonclinical-rule, abbvie-iambic-ai-drug, and iambic-enchant-v3.