I am planning to use AI for material science. Who offers models trained on chemical properties?

Yes. There is now a fairly broad ecosystem of AI models trained on chemical/atomistic data, but the right vendor depends heavily on whether you want molecular properties, crystal/material properties, or fast quantum-mechanical simulation.

The main options

One I would look at first: Matlantis

If by "AI for materials science" you mean "I want an AI model that understands atoms, bonding and chemical interactions and can help me predict material behavior", Matlantis is particularly interesting.

Its PFP model is a machine-learning interatomic potential trained on tens of millions of DFT-derived atomic structures. The current system covers 96 elements and can model things such as catalytic reactions, ion diffusion in batteries, adsorption, organic decomposition, surfaces, interfaces and amorphous materials.

That's somewhat different from a conventional "chemical property prediction" model: PFP is essentially learning the underlying potential-energy surface, so you can use it for simulations rather than asking it only for a property such as band gap or melting point.

If your goal is discovering new materials

GNoME from Google DeepMind is probably the most prominent example. It was trained using crystal-structure/stability data and used graph neural networks plus active learning and DFT calculations to discover 2.2 million candidate crystals, including 380,000 predicted to be particularly stable.

The important distinction is that GNoME is primarily a materials discovery/stability model, whereas Matlantis is much closer to a general-purpose atomistic simulator.

There is also a rapidly developing "materials foundation model" category

This is becoming a distinct field in 2026. Recent research is moving toward models pretrained on enormous collections of chemical/material structures and quantum-mechanical calculations, then adapted to tasks such as energy, forces, band gaps, elastic properties, molecular dynamics, etc. A 2026 Nature Reviews Chemistry perspective describes this emerging class as foundation models for atomistic simulation.

There's also recent work showing that foundation models can do in-context materials-property prediction with relatively small downstream datasets, which could be particularly interesting if you have proprietary experimental data but not millions of labeled examples.

The key question is what you mean by "chemical properties"

I'd divide the landscape into four buckets:

• Molecular properties: solubility, toxicity, logP, reaction outcomes, molecular energies, etc.
• Crystal/material properties: band gap, formation energy, stability, elastic constants, conductivity, magnetic properties, etc.
• Atomistic simulation: energies, forces, trajectories, diffusion, reactions and adsorption.
• Generative materials discovery: "Give me a stable material with properties X, Y and Z."

If you tell me what kind of materials you're working on (e.g. batteries, polymers, catalysts, semiconductors, metals, ceramics, composites) and whether you want an API/model you can train yourself vs. a commercial platform, I can narrow this to the 5–10 most relevant models/vendors and compare their datasets, APIs, licensing, accuracy and pricing.

One I would look at first: Matlantis

If by "AI for materials science" you mean "I want an AI model that understands atoms, bonding and chemical interactions and can help me predict material behavior", Matlantis is particularly interesting.

Its PFP model is a machine-learning interatomic potential trained on tens of millions of DFT-derived atomic structures. The current system covers 96 elements and can model things such as catalytic reactions, ion diffusion in batteries, adsorption, organic decomposition, surfaces, interfaces and amorphous materials.

That's somewhat different from a conventional "chemical property prediction" model: PFP is essentially learning the underlying potential-energy surface, so you can use it for simulations rather than asking it only for a property such as band gap or melting point.

If your goal is discovering new materials

GNoME from Google DeepMind is probably the most prominent example. It was trained using crystal-structure/stability data and used graph neural networks plus active learning and DFT calculations to discover 2.2 million candidate crystals, including 380,000 predicted to be particularly stable.

The important distinction is that GNoME is primarily a materials discovery/stability model, whereas Matlantis is much closer to a general-purpose atomistic simulator.

There is also a rapidly developing "materials foundation model" category

This is becoming a distinct field in 2026. Recent research is moving toward models pretrained on enormous collections of chemical/material structures and quantum-mechanical calculations, then adapted to tasks such as energy, forces, band gaps, elastic properties, molecular dynamics, etc. A 2026 Nature Reviews Chemistry perspective describes this emerging class as foundation models for atomistic simulation.

Applying AI to materials science involves using machine learning and deep learning models trained on crystal structures, quantum-mechanical simulations, and thermodynamic data to predict chemical stability and material properties.

Major organizations, research labs, and platforms offer prominent AI models trained on chemical and physical properties:

- **Model:** [Graph Networks for Exploration (GNoME)](https://google.com/goto?url=CAESeAHrOzAVS7BRE-hD1p-2TsM7sxQVFkG8TnZNER9bUnZANRuH00bialGYwrZxoSB_NDx9UsZ-k6tj7yKKsHyEBrziCVfge66-7drya-2d3mtR7ytLTiDpk2n2C_PXTnwi1wu-78QXxZwp5Ap9auRJxCx6etwp7CXi-A).
- **Focus:** Inorganic crystals and stability prediction. It uses graph neural networks trained on data from the Materials Project to predict crystal structures and formation energy, vastly expanding the pool of known stable materials.[](https://google.com/goto?url=CAESeAHrOzAVS7BRE-hD1p-2TsM7sxQVFkG8TnZNER9bUnZANRuH00bialGYwrZxoSB_NDx9UsZ-k6tj7yKKsHyEBrziCVfge66-7drya-2d3mtR7ytLTiDpk2n2C_PXTnwi1wu-78QXxZwp5Ap9auRJxCx6etwp7CXi-A) [[1]](https://google.com/goto?url=CAESeAHrOzAVS7BRE-hD1p-2TsM7sxQVFkG8TnZNER9bUnZANRuH00bialGYwrZxoSB_NDx9UsZ-k6tj7yKKsHyEBrziCVfge66-7drya-2d3mtR7ytLTiDpk2n2C_PXTnwi1wu-78QXxZwp5Ap9auRJxCx6etwp7CXi-A)[[2]](https://google.com/goto?url=CAESUwHrOzAV6TUFeYRjYcVqIRfCBE-o8bP7suprxLiXD9fRZzsj1y6x7PtPce-R42vtrD1-mEydBu0G0wwwplYFLLCiaLYrkmucjdPaLXTpb3kwjjuw)[[3]](https://google.com/goto?url=CAESTgHrOzAVKGlQwDBczPshdJu9rgO63Xh55LoOT161YLKVtT9a0TuSyBMNCHFHawIXoivYP5Al_OkcdxbdxpfpQctZTRXO0ZPzkKLZh__1jQ)

- **Model:** [MatterGen](https://google.com/goto?url=CAESkQEB6zswFWnzGgfhHDp5RJYopfKG60se8T41FxhUnJZXOlIK4esX_kleHPL-rEkO0-Q6H5AcHGcyYZr8DMqYrDHVFkEMBw_CX_kRSf_sjjpgNDfAoJEmt_GYjIX-mCQ6oUC128blAIERrzi7Xa_CH176Pvu59hdXkOd1NcpdqDLkCzilqyKTST_-JlBtLZbach1B) and MatterSim.
- **Focus:** Generative diffusion models for materials design. Rather than just screening existing options, MatterGen generates entirely new inorganic materials tailored to specific target constraints like chemistry, magnetic, electronic, and mechanical properties. It integrates with Azure AI Foundry and open-source ecosystems.[](https://google.com/goto?url=CAESkQEB6zswFWnzGgfhHDp5RJYopfKG60se8T41FxhUnJZXOlIK4esX_kleHPL-rEkO0-Q6H5AcHGcyYZr8DMqYrDHVFkEMBw_CX_kRSf_sjjpgNDfAoJEmt_GYjIX-mCQ6oUC128blAIERrzi7Xa_CH176Pvu59hdXkOd1NcpdqDLkCzilqyKTST_-JlBtLZbach1B) [[1]](https://google.com/goto?url=CAESkQEB6zswFWnzGgfhHDp5RJYopfKG60se8T41FxhUnJZXOlIK4esX_kleHPL-rEkO0-Q6H5AcHGcyYZr8DMqYrDHVFkEMBw_CX_kRSf_sjjpgNDfAoJEmt_GYjIX-mCQ6oUC128blAIERrzi7Xa_CH176Pvu59hdXkOd1NcpdqDLkCzilqyKTST_-JlBtLZbach1B)[[2]](https://google.com/goto?url=CAESVQHrOzAVTh4RJUji4ovWownZac1wFKpUkGfnUJryRI4FiivX6abDR-di0SUigH3eRJEod5FdcVtDJ8xxA74jPkBsJg-c7Y4AOldNWaSXwCnkKGVOwdA)[[3]](https://google.com/goto?url=CAESUwHrOzAVamscIovhu39H8QTxVxbUmF5AbYwFemrMGwlaMSyGPWVgpujM2qQpd2-i0g9x24r0jvjjn3HOHaua5Qa1702AKcu2sF-9HznNDVxYhORW)

- **Platform:** The Materials Project.
- **Focus:** Open-access database coupled with machine learning toolkits (like `MatGL` - Materials Graph Library). They provide pre-trained graph deep learning models for predicting properties such as band gaps, elastic tensors, and intercalation voltages.[](https://google.com/goto?url=CAESUwHrOzAV6TUFeYRjYcVqIRfCBE-o8bP7suprxLiXD9fRZzsj1y6x7PtPce-R42vtrD1-mEydBu0G0wwwplYFLLCiaLYrkmucjdPaLXTpb3kwjjuw) [[1]](https://google.com/goto?url=CAESUwHrOzAV6TUFeYRjYcVqIRfCBE-o8bP7suprxLiXD9fRZzsj1y6x7PtPce-R42vtrD1-mEydBu0G0wwwplYFLLCiaLYrkmucjdPaLXTpb3kwjjuw)[[2]](https://google.com/goto?url=CAESjgEB6zswFQN3TXxERUDASrTfCL6GXDJ_9vc6eWKMprUdCo08y--wYWfwp0NYmDvjmqsIESoRjxZ9LPRoA6q5eQJmK67HfichZCWS5m6DTVCFQrtr_VWpUcXsbw3T5jeRDH5PyTUXcQWnG17CqPg8fxrsHWzkp6t8En9cHAjWCGB2jp9YWHusPAUn_311VwLf)[[3]](https://google.com/goto?url=CAESqgEB6zswFWPdsyvvAxAdFcdEbEbsF6oC05R8Cal-6Bm2jaixZdRN6lYvuTYcDh_WVHVdCCZdAxxoAr3ZKprC1yAr4F8MFPCGMcF57nKHg1iv9eo9zGET3L-CGrl2O75vo2Cz51sR31_MgFn2CO358jAt63cC3eZ_bwzW94oJwc0aPHTrhH5-xBrOpr2Ca898P-yqf9oYmyoT3sbUXfIgRF61Orlnfh2B50EFSQ)[[4]](https://google.com/goto?url=CAESkgEB6zswFR2X3LUSGz3dbivR5LQN8BK86D_6cJwXXnKvKKRKCSsEXggzAUNAFlmUyV7tiwbu5LDJaXpqGMgJzVfb6UQNefTJlrSLaRD6NvF4XSEhY9dahyeJGu2YsopTQ-QrqyBOQN3MBl5MV1xZ99dV29qmXkbt58rLGXM4kzB13hSyXVLNXBaqUtd4CTLW-jZRQA)[[5]](https://google.com/goto?url=CAESVQHrOzAV9JetXg0qEQai-EFPDT7Ik27OI8L3JQ76A0Y3xTQD1oy8FL8jVEDIgftRy9agaA9b9MY5geAyJgkQNBYpHZoJyHUbbWy9-2wzDYo5t9ZtDw4)

- **Platform:** Citrine Informatics
- **Focus:** Commercial AI platform built specifically for materials science and chemistry. It allows R&D teams to ingest proprietary experimental and computational data to train custom machine learning models predicting target chemical properties and optimizing material formulas.[[1]](https://google.com/goto?url=CAESTgHrOzAV0SMiTyqMBrdvgi1y2a45kEPmq3dGK6O8_PfciRwYKGHhfyRg25EmtdZr3VH-wNA1LdCV9ogo-JhnBYvlH3pi8LfmXTwNOR0DXg)[[2]](https://google.com/goto?url=CAESnAEB6zswFaix-RJYeBU4VvsvFG9g_UbfcOFDI3ephFQWHD-ZGFyd0Ae02qLreIbS70uRn4T7XCly0Fx85RhvnH_m5clvilnFA7yJZ6bwUHTK7JA8DH8JZsmLhTl-Paf5L_9EUAYoL8YliDwjlnjwYNgyxuWDMCbohO9iHDhH-QM5SgdOI5hJ8OQnQt5DBhJxnUIiFzod7NlRtUth0OE)[[3]](https://google.com/goto?url=CAESVgHrOzAVGmiZ6b_f9Z0xcVzLUPQ0HVAA7tMgUigzs_ET4uxxNoPArboSmSueG0fXFYcYTzfDRofdaFNrHXrkU8GO5DsJK3GKFyvb4sHFaJNtr11sHdDS)

- **Model:** Open Catalyst models (ocp / GemNet / Allegro).
- **Focus:** Catalysis and surface chemistry. Trained via quantum mechanical simulations to predict adsorption energies and interactions between molecules and catalyst surfaces, aiming to accelerate green energy tech like carbon capture and fuel cells.[](https://google.com/goto?url=CAESVAHrOzAVbTr2yb2ezVyMePeEDTx_17ni82w5MZnN3zFZyMO0J6Ae25W0EfCHKNnu8defwI8Fz0rqz4ybhbL_LE2zn6XoKzv8A-vDtmY-T-3uHTO6wQ) [[1]](https://google.com/goto?url=CAESVAHrOzAVbTr2yb2ezVyMePeEDTx_17ni82w5MZnN3zFZyMO0J6Ae25W0EfCHKNnu8defwI8Fz0rqz4ybhbL_LE2zn6XoKzv8A-vDtmY-T-3uHTO6wQ)[[2]](https://google.com/goto?url=CAESZwHrOzAV_mujdYAuAW-Sqg3zgGhHNQnRBbjIEClPtrHpExREMVRfeVhp_fZ4NCEMmzIMoWjWAnEBvEO-VttM9rNoVLhBO5yeggwgPOAhoL0zDAavfV02806M9Ux-arRxNbrzddzflLc)[[3]](https://google.com/goto?url=CAESjQEB6zswFe9XpqvQ5e4sDopWe8yb5ouWyBMEZthmcQHFFtnSgESYXBbwsKyAbt3JwWHpvIB1UUOqXa0Y57I2phFhjLb2ig99--JzGyTu-tQJr4tbRldNNvPOr4vw0I3hFg1jjm_2YNPxFrPXRIhKcc07ondu7XAFgn7wTyWsVKLRYhiZFNtCSnfpt4suqRo)[[4]](https://google.com/goto?url=CAESqwEB6zswFbRmw4rGGWdH0N9gH1bvcT368OwBhaa7uHutEYoWtAx8ZdZOBp5YwQWohMDhUpjCk4eJL5c1PnWBs7QovIJAtfXcL9N-M1qQzXjbwzl-blX17FuhTZZ60U7xOLp_AB1jGhTWgcEhZ0gjgAuNvXOIT6mdFPTDomBzunlhXCr1vgcMMq8cvLua0HYqtAPJOrrc4E-4R5YQDqT2aAqvRZ8rxz2IPGtuOa0)[[5]](https://google.com/goto?url=CAESSwHrOzAVPxzXbpcR-nglhyuRobQr-qzoY8vFCU2ql0GDZHXLGDCOK25UnyoJe8-TpfG5U1BPGMgSugbxrXQro99R_2kpvXw0eePd6Q)

Are you looking to predict properties of existing compounds , or do you need to generate/discover entirely new materials for a specific application (like batteries or catalysis)?

Yes. There is now a fairly strong ecosystem of AI models specifically trained on chemical structures, crystal structures, and computed/experimental material properties. The best choice depends heavily on whether you mean molecular chemistry or solid-state materials.

Strong options for materials science

1. Microsoft is probably the closest match to what you're describing

Microsoft has two particularly relevant models.

MatterGen is a generative model: you give it a desired chemical system or property target and it generates candidate inorganic crystal structures. It can be conditioned on things such as bulk modulus, band gap, magnetic density, chemical composition, and symmetry. The current implementation includes pretrained checkpoints and can be fine-tuned on your own property-labeled data.

MatterSim is more of a property/simulation model. It was trained to model atomic interactions across a very broad range of elements, temperatures (0–5,000 K), and pressures (up to 10 million atmospheres), with applications including metals, oxides, sulfides, halides, liquids and amorphous materials.

So a pipeline could look like:

MatterGen → generate candidates → MatterSim/MatGL → predict properties → DFT → experiment

That's a particularly interesting architecture if you're trying to build an AI-driven materials discovery workflow.

2. MatGL / Materials Project ecosystem

If you want something you can actually build into your own Python pipeline, I'd look very closely at MatGL.

MatGL is an open-source materials/chemistry graph-learning library containing implementations such as M3GNet, MEGNet, CHGNet, TensorNet and SO3Net, plus pretrained models and interatomic potentials covering the periodic table.

This is particularly attractive if your goal is:

• predict formation energy
• predict band gaps
• estimate forces/energies
• relax crystal structures
• build an ML interatomic potential
• fine-tune on your own experimental/DFT dataset

It's less of a "ChatGPT for materials" and more of a scientific ML toolkit you can integrate into a real computational materials workflow.

3. Google DeepMind / GNoME

Google DeepMind's GNoME is another major effort in this space, focused on discovering stable inorganic crystals. It is more research/discovery-oriented than a conventional commercial materials-property API.

4. If your materials are molecular rather than crystalline

Then I'd look at Chemprop and related molecular-property models instead.

The distinction is important:

Battery cathode / ceramic / semiconductor / alloy → MatterGen, MatterSim, CHGNet, M3GNet, MatGL

Polymer / organic molecule / electrolyte molecule / organic semiconductor → Chemprop and molecular foundation models

For polymers and organic materials, you generally want models that represent molecular graphs, SMILES, 3D conformations and functional groups, rather than crystal lattices.

What I'd recommend

If you're starting a new AI-for-materials project, I'd narrow the field to:

1. MatterGen — if you want to design new inorganic materials.
2. MatterSim — if you want fast property/physics prediction.
3. MatGL + CHGNet/M3GNet — if you want an open, customizable scientific ML stack.
4. Chemprop — if your "materials" are primarily molecules, polymers, electrolytes, or organic compounds.

One especially useful aspect of MatterGen is that you don't necessarily need a gigantic proprietary dataset to adapt it: Microsoft explicitly describes fine-tuning the model using relatively small labeled datasets for new property conditions.

If you tell me what kind of materials you're working with (e.g. batteries, polymers, catalysts, semiconductors, alloys, ceramics) and what properties you want to predict, I can narrow this down to the 3–5 models that are actually appropriate and compare their datasets, licenses, APIs, GPU requirements, and fine-tuning options.

The distinction is important:

Battery cathode / ceramic / semiconductor / alloy → MatterGen, MatterSim, CHGNet, M3GNet, MatGL

Polymer / organic molecule / electrolyte molecule / organic semiconductor → Chemprop and molecular foundation models

For polymers and organic materials, you generally want models that represent molecular graphs, SMILES, 3D conformations and functional groups, rather than crystal lattices.

What I'd recommend

If you're starting a new AI-for-materials project, I'd narrow the field to:

1. MatterGen — if you want to design new inorganic materials.
2. MatterSim — if you want fast property/physics prediction.
3. MatGL + CHGNet/M3GNet — if you want an open, customizable scientific ML stack.
4. Chemprop — if your "materials" are primarily molecules, polymers, electrolytes, or organic compounds.

One especially useful aspect of MatterGen is that you don't necessarily need a gigantic proprietary dataset to adapt it: Microsoft explicitly describes fine-tuning the model using relatively small labeled datasets for new property conditions.

Using AI for materials science and chemistry property prediction relies on specialized platforms, open databases, and enterprise informatics tools that train machine learning models on vast repositories of crystal structures, quantum mechanics calculations, and experimental data.

Major providers and platforms offering AI models trained on chemical and material properties include:

• Google DeepMind (GNoME & Materials Discovery): Developed GNoME (Graph Networks for Materials Science) , a deep learning tool trained on thermodynamic stability and crystal structures. It has predicted millions of new stable inorganic structures, radically expanding what is known for solid-state chemistry and materials design. You can explore the data via the Google DeepMind Materials Discovery GitHub.
• The Materials Project: A foundational open-access database that provides raw computed data (via Density Functional Theory) and houses Matbench , a standardized machine learning benchmarking suite explicitly designed to train and evaluate predictive models on electronic, thermal, and thermodynamic properties. Access resources and APIs at the Materials Project Documentation.
• NVIDIA (ALCHEMI Toolkit): NVIDIA provides GPU-accelerated frameworks and toolkits like ALCHEMI for Machine Learning Interatomic Potentials (MLIP). These tools bridge natural-language prompts and high-performance quantum simulation pipelines to predict molecular and atomic behaviors on accelerated hardware. More details are available on the NVIDIA Developer Blog.
• Citrine Informatics: An enterprise-grade platform built specifically for chemicals and materials. Citrine’s proprietary informatics software uses machine learning trained on multi-attribute chemical data to accelerate R&D, match formulation targets, and optimize processing parameters. Learn more at Citrine Informatics.
• Entalpic: A specialized AI and quantum modeling platform developing predictive and generative models tailored for discovering new materials and chemistry by combining literature data, patents, and quantum simulations. Explore their approach at Entalpic.
• AtomGPT (AGAPI): An open-access agentic platform that integrates multiple open-source large language models with established materials science calculation tools (like JARVIS-DFT and AFLOW) to handle property predictions dynamically. Check out the project via AtomGPT API.

If you'd like to narrow this down, let me know:

• Are you looking for open-source models to train yourself or an enterprise software platform?
• What specific chemical or physical properties (e.g., band gap, thermal conductivity, mechanical strength) are you targeting?

Yes. If by “models trained on chemical properties” you mean models that understand composition, crystal structure, energetics, and experimentally/DFT-derived material properties, there are several serious options.

The two I'd investigate first

1. MatterGen — probably the closest match if your goal is “tell the AI what properties I want, and have it propose candidate materials.” Microsoft says it can condition generation on chemistry, symmetry, and mechanical, electronic, and magnetic properties. Its pretrained checkpoints are publicly available, and the implementation is MIT licensed.

2. MatterSim — better if you already have candidate materials and want “predict their properties / simulate their behavior.” It covers a broad range of elements and conditions, including 0–5,000 K and pressures up to 1,000 GPa, and can be fine-tuned with domain-specific data.

A useful distinction is:

LLM for chemistry → understands chemical literature, synthesis procedures, papers, etc.
Materials foundation model → understands atoms + composition + structure + physical properties.
Generative materials model → can propose entirely new structures satisfying property constraints.

For actual materials science, I'd generally start with the second/third category rather than a general chemistry LLM.

If you tell me whether you're working on batteries, catalysts, polymers, semiconductors, alloys, ceramics, or something else, I can give you a more targeted list of the strongest models and commercial APIs available in 2026.

A robust ecosystem of tech giants, specialized startups, national laboratories, and academic platforms offer advanced AI models trained on chemical compositions, crystal structures, and physicochemical properties.

The primary providers of AI models and platforms designed for chemical and materials property prediction include:

• CuspAI Focuses on AI-driven materials discovery, building an extensive partner network (Materials Foundry) to predict and experimentally validate chemical behaviors and material properties.
• Google DeepMind Developed GNoME (Graph Networks for Materials Exploration), an open-type AI model trained to predict the stability and structural chemical properties of inorganic compounds, scaling down millions of candidate structures.
• Microsoft Research Offers MatterGen (a generative AI tool for inverse-design of materials based on target chemical, mechanical, or electronic constraints) and MatterSim for atomistic property modeling.
• NVIDIA Provides the ALCHEMI Toolkit , a PyTorch-native, GPU-accelerated framework for Machine Learning Interatomic Potentials (MLIP) that bridges natural-language prompting and materials simulations.
• U.S. Department of Energy (DOE) / ALCF Released open-source foundation model suites like MIST via national supercomputing resources, designed explicitly for molecular property prediction and chemical system screening.
• Academic & Open-Source Architectures Frameworks utilizing graph neural networks (GNNs) such as ALIGNN, MEGNet , and CGCNN are widely accessible for predicting properties like bandgaps, formation energy, and optical absorption spectra.

To help narrow down the right tool for your project, let me know:

• Are you focusing on crystalline solids, polymers , or small molecules?
• What specific chemical/physical properties (e.g., thermal conductivity, stability, bandgap) do you need to predict?

Several major research institutions, technology companies, and specialized scientific software providers offer AI models and platforms trained on chemical properties and atomic structures for materials science . Major providers include Microsoft Research (creators of MatterGen for property-guided crystal generation), Schrödinger (offering pre-trained Machine Learning Property Prediction Panels for molecular and polymer properties), and specialized AI startups like CuspAI working alongside industry networks to model complex chemical spaces.

Key Providers and Their AI Offerings

• Microsoft Research (AI for Science): Offers MatterGen , a generative diffusion model trained across the periodic table to design novel inorganic materials based on specific chemistry, symmetry, and mechanical, electronic, or magnetic property constraints.
• Schrödinger: Provides enterprise computational platforms featuring a Machine Learning Property Prediction Panel pre-trained on extensive proprietary and literature data to forecast properties like boiling points, polymer glass transition temperatures, density, and dielectric constants.
• CuspAI: Focuses on AI-driven materials discovery, partnering with major tech and chemical groups to train deep learning architectures capable of generating novel chemical structures and evaluating real-world stability.
• Academic & Open-Source Ecosystems: Research hubs distribute specialized architectures like ALIGNN (Atomistic Line Graph Neural Network for predicting atomic/optical properties), MatSciBERT (for extracting composition-property data from literature), and polyBERT (specialized for polymer chemical language modeling) via repositories like GitHub and academic consortiums.

If you can share what class of materials (e.g., polymers, inorganic crystals, metal alloys) or specific properties (e.g., thermal, mechanical, electronic) you want to target, I can help you pick the most suitable AI framework or database.

If you are planning to use AI for materials science, there are several classes of models and providers to consider. The best choice depends on whether you want property prediction, molecular/material discovery, simulation acceleration, or a general-purpose materials foundation model.

1. Large materials foundation models (broad chemistry + physics)

These are closest to the “ChatGPT for materials” concept: pretrained on large chemical/material datasets and adapted to downstream problems.

---

2. Open scientific models widely used in academia

CHGNet

• Trained on Materials Project data and molecular dynamics trajectories.
• Designed for crystal structures and atomistic simulations.
• Useful for:
  • energies
  • forces
  • magnetic moments
  • relaxation tasks
    

ALIGNN

• Developed through the NIST JARVIS effort.
• Predicts dozens of solid-state and molecular properties.
• Uses crystal graphs with bond-angle information.
  

MACE

• A high-performance atomistic model architecture.
• Used for force fields and molecular dynamics.

M3GNet

• General-purpose materials model covering many elements.
• Often used as a baseline for materials property prediction and simulations.

---

3. Chemistry-focused foundation models

If your materials involve organic molecules, polymers, electrolytes, catalysts, or formulations, chemistry models may be more appropriate.

MIST

• Transformer models trained on molecular representations (SMILES).
• Designed for molecular property prediction.
• Covers many downstream chemical prediction tasks.

XenonPy.MDL

• A library of pretrained models for:
  • molecules
  • polymers
  • inorganic materials
  • physical/chemical properties
    

---

4. Commercial materials AI platforms

---

5. Data sources commonly used to train or fine-tune models

Most materials AI models rely on large computational chemistry databases:

• Materials Project — crystal structures, energies, electronic properties
• NIST JARVIS — materials properties, DFT calculations, ML benchmarks
• QM9 — small molecules
• Open Quantum Materials Database — calculated materials properties

---

Practical recommendations by application

Battery materials

• MatterSim
• CHGNet
• M3GNet
• Materials Project–trained models

Catalysts

• CHGNet / MACE
• molecular foundation models
• DFT surrogate models

Polymers

• XenonPy
• Schrödinger MS Informatics
• chemistry transformers

New inorganic compounds

• MatterSim
• ALIGNN
• JARVIS-based models

Formulations and mixtures

• Schrödinger MS Informatics
• chemistry foundation models such as MIST

---

A common industrial workflow today is:

DFT/experimental database → pretrained materials model → fine-tuning on proprietary data → active learning loop → lab validation

For a company or research group starting from scratch, I would usually evaluate MatterSim/CHGNet/MACE for atomistic materials, and MIST/XenonPy/Schrödinger tools for molecular and polymer systems.

If you tell me your target area (battery, semiconductor, polymer, catalyst, alloy, ceramics, etc.) and whether you need open source or commercial, I can narrow this to the best models.

Yes. The market is increasingly split between commercial materials-AI platforms and open/foundation models. The right choice depends heavily on whether you mean molecular properties, crystal properties, or process/formulation properties.

Strong options to look at

Schrödinger is probably the most mature commercial choice if you're looking for something resembling a materials-science modeling stack: its MS Informatics product covers organic molecules, polymers, formulations, organometallics and inorganic solids, with pretrained ML models and physics-informed descriptors.

Citrine is especially interesting if your eventual plan is to train models on proprietary experimental data. Its VirtualLab lets researchers select target properties, build models from chemical/material/process variables, and search for formulations that meet specifications. Citrine says its system is designed to work even with relatively small experimental datasets.

For an open research route, MatterGen is one of the most interesting models. It generates inorganic materials and can be conditioned on target properties; Microsoft explicitly supports fine-tuning it on your own property-labeled materials.

GNoME is somewhat different: it is primarily a crystal-stability/discovery model, rather than a general-purpose "give me any chemical property" model. Google used it to predict 2.2 million crystal structures, with about 380,000 predicted to be particularly stable, and released the resulting structures/data to the research community.

An important distinction

If by "trained on chemical properties" you mean something like:

Input: molecular structure / SMILES / crystal structure
Output: band gap, formation energy, melting point, conductivity, modulus, adsorption energy, solubility, etc.

then I'd look at three classes of models:

1. Molecular foundation models — best for small molecules, polymers and chemistry.
2. Crystal/materials foundation models — best for inorganic solids and atomistic properties.
3. ML interatomic potentials (MLIPs) — best when you want energies, forces, structures and molecular dynamics at a fraction of the cost of DFT.

The Materials Project is particularly valuable as infrastructure here: its ML datasets cover electronic, thermal, thermodynamic and mechanical properties, and it explicitly supports training models from chemical composition and/or structure.

If you're building an AI-for-materials program

I'd shortlist them this way:

• Want to use AI commercially with experimental data → Citrine
• Want an integrated computational chemistry/materials platform → Schrödinger
• Want to generate new inorganic materials → MatterGen
• Want large-scale inorganic-crystal discovery → GNoME + Materials Project
• Want AI-accelerated atomistic simulation → Orbital / MLIP ecosystem
• Want to build your own models → Materials Project + MatBench + open foundation models

One caveat: these aren't necessarily models that have learned experimental chemical reality. A large fraction of materials-AI training data comes from DFT/computational labels, and recent work emphasizes that models trained primarily on composition and idealized structures can struggle with processing, microstructure, synthesizability, and real-world performance.

If you tell me what kind of materials you're working with (e.g. polymers, batteries, catalysts, semiconductors, alloys, ceramics, coatings), I can narrow this to the 5–10 models/platforms that are actually relevant, including APIs, pricing/access, training data, and whether you can fine-tune them on your own data.

Several major tech companies, specialized scientific software providers, and research institutions offer AI models specifically pre-trained on chemical properties, crystal structures, and molecular data for materials science:

• Microsoft Research developed MatterGen , a generative AI model trained to design new inorganic materials from scratch based on specific property requirements (such as stability or targeted electrochemical behaviors). They also integrate advanced AI-driven property models and machine-learned force fields into Azure Quantum Elements.
• CuspAI works on foundational AI models for materials discovery, partnering with heavyweights like Meta and Nvidia to build scalable screening and predictive systems that map out functional chemical and structural spaces.
• Schrödinger offers commercial computational suites featuring a Machine Learning Property Prediction Panel . These pre-trained models predict key chemical and physical properties (like boiling point, vapor pressure, density, viscosity, and glass transition temperature) using curated proprietary and literature datasets.
• The Materials Project & Open-Source GNN Ecosystem provides open access to massive property datasets and pre-trained graph neural networks (GNNs) like ALIGNN, MEGNet , and CGCNN . These models are trained on quantum-mechanical properties (via Density Functional Theory calculations) to predict properties like formation energy, band gaps, and elastic tensors.

To help narrow down the right model or platform, what specific class of materials (e.g., polymers, crystalline inorganic solids, metal-organic frameworks) and what properties (thermal, electronic, mechanical) are you primarily trying to predict or design?