SandboxAQ

SandboxAQ builds what it calls Large Quantitative Models: AI systems trained on physics, chemistry and proprietary scientific datasets rather than on general text. The company's thesis is that LLMs excel at generating content while quantitative work — predicting a molecule's behaviour, a material's properties or a network's cryptographic risk — requires models grounded in real-world science. Its models are exposed through a common platform, REAQT, which combines an LLM interface with LQM modules, proprietary datasets and compute orchestration.

The applied portfolio spans several domains. BioSim covers virtual screening, antibody design, lead optimization, toxicity prediction and biomarker development. ChemSim addresses catalysis, batteries, alloys and property prediction. Beyond simulation, SandboxAQ ships agentic products and security tooling, including AQtive Guard for post-quantum cryptography management and cryptographic discovery, plus a medical diagnostics line, AQMed, and AQNav, a quantum-sensing approach to GPS-denied navigation. The company is also known for collaborations with major AI labs to make its models reachable through familiar LLM interfaces.

Commercial engagement is enterprise-first: drug discovery, chemicals and materials, financial services, defense and healthcare, with engagements sold through demos, pilots and direct conversations rather than self-serve plans. There is no published price list, and public signals such as a large CHIPS R&D award underline that this is a research-heavy, government- and enterprise-facing company rather than a consumer product.