Large Quantitative Models (LQMs) are a fundamentally different category of AI from the Large Language Models (LLMs) that power chatbots like ChatGPT and Gemini. According to SandboxAQ, LQMs are "AI anchored in the physical world" .

How LQMs Differ from LLMs

The differences fall into three areas: training data, objective, and output.

Data and training. LLMs are trained on internet text and natural language — books, articles, and conversations — to understand and generate human language. LQMs, by contrast, are trained on quantitative data and the fundamental laws of physics, chemistry, and mathematics . SandboxAQ describes them as built on numerical structured data rather than text .

Objective. LLMs optimize for language understanding and generation — predicting the next word in a sentence or summarizing a document. LQMs optimize for specific quantitative objectives, such as predicting material properties, chemical behavior, or financial risk metrics .

Output. LLMs produce text, summaries, and conversation. LQMs produce predictions, simulations, and engineering insights grounded in physical reality . As SandboxAQ explains, where LLMs can describe a chemical reaction, LQMs can simulate it .

SandboxAQ CEO Jack Hidary has said LQMs will eclipse LLMs in verticals like pharma, materials, energy, and finance . The company emphasizes that LQMs are not replacements for LLMs but complementary tools — LLMs handle human interaction while LQMs handle quantitative precision .

What Models Are Being Offered?

SandboxAQ is initially offering two LQM products through the Google Cloud Marketplace :

• AQCat (materials discovery) — will be available first
• AQPotency (drug discovery) — will follow after AQCat

Both are accessible through the marketplace. SandboxAQ has also made its LQMs accessible via leading LLMs, so researchers can query them through natural language interfaces they already know .

What Is the Gemini for Science Bundle?

Gemini for Science is not a single model. It is a suite of experimental AI tools announced by Google at I/O 2026 (May 19–20, 2026), designed to accelerate the scientific method . It is best understood as an "orchestration layer" for scientific research .

Gemini for Science bundles three core pillars:

• Hypothesis Generation — built on Co-Scientist, a team of AI agents that work together to generate, debate, and evaluate research hypotheses. Google published a same-day Nature paper validating the approach .
• Computational Discovery — built on AlphaEvolve and Empirical Research Assistance (ERA). ERA reportedly beat the CDC's forecasting model for infectious disease prediction in benchmarks published at launch .
• Literature Insights — built on NotebookLM, providing automated literature summarization and research analysis .

Additionally, Gemini for Science includes Science Skills, a separate layer in Google Antigravity that connects to over 30 life-science databases .

The experimental tools are available through Google Labs at labs.google/science, with gradual access beginning at launch .

Commercial Logic Behind the Partnership

For Google Cloud: SandboxAQ's LQMs represent a new, high-value workload category — scientific and quantitative AI — that drives cloud consumption. Google Cloud gains a differentiated, enterprise-grade product alongside its general-purpose Gemini models, positioning itself as the infrastructure provider for "AI-native scientific discovery" .

For SandboxAQ: The Google Cloud Marketplace provides immediate access to Google's vast enterprise customer base, a trusted billing and procurement channel, and scalable infrastructure — including integration with NVIDIA DGX Cloud on Google Cloud . SandboxAQ, an Alphabet spin-out now valued at $5 billion, deepens a natural commercial relationship .

What the Marketplace Listing Means for Scientific AI

This is a commercial milestone for scientific AI. It marks the first time specialized Large Quantitative Models — a category distinct from general-purpose LLMs — are available through a major cloud marketplace with enterprise procurement, billing, and security .

It signals that scientific AI is moving from research labs and custom deployments into a standardized, commercially listed product category that enterprises can buy off the shelf. SandboxAQ CEO Jack Hidary has said LQMs will eclipse LLMs in verticals like pharma, materials, energy, and finance , and the marketplace availability is the infrastructure that makes that shift real.

Key Takeaways

• LQMs are trained on physics, chemistry, and math — not text — and optimize for quantitative predictions rather than language generation.
• AQCat (materials) and AQPotency (drugs) are the first two LQM models available on Google Cloud Marketplace.
• Gemini for Science is an orchestration suite, not a single model, bundling Co-Scientist, AlphaEvolve, ERA, and NotebookLM.
• The partnership positions Google Cloud as the infrastructure backbone for scientific AI while giving SandboxAQ enterprise distribution.
• The marketplace listing represents the first time specialized LQMs are available through an enterprise cloud procurement channel.

Large Quantitative Models (LQMs) are a fundamentally different category of AI from the Large Language Models (LLMs) that power chatbots like ChatGPT and Gemini. According to SandboxAQ, LQMs are "AI anchored in the physical world" .

How LQMs Differ from LLMs

The differences fall into three areas: training data, objective, and output.

Data and training. LLMs are trained on internet text and natural language — books, articles, and conversations — to understand and generate human language. LQMs, by contrast, are trained on quantitative data and the fundamental laws of physics, chemistry, and mathematics . SandboxAQ describes them as built on numerical structured data rather than text .

Objective. LLMs optimize for language understanding and generation — predicting the next word in a sentence or summarizing a document. LQMs optimize for specific quantitative objectives, such as predicting material properties, chemical behavior, or financial risk metrics .

Output. LLMs produce text, summaries, and conversation. LQMs produce predictions, simulations, and engineering insights grounded in physical reality . As SandboxAQ explains, where LLMs can describe a chemical reaction, LQMs can simulate it .

SandboxAQ CEO Jack Hidary has said LQMs will eclipse LLMs in verticals like pharma, materials, energy, and finance . The company emphasizes that LQMs are not replacements for LLMs but complementary tools — LLMs handle human interaction while LQMs handle quantitative precision .

What Models Are Being Offered?

SandboxAQ is initially offering two LQM products through the Google Cloud Marketplace :

• AQCat (materials discovery) — will be available first
• AQPotency (drug discovery) — will follow after AQCat

Both are accessible through the marketplace. SandboxAQ has also made its LQMs accessible via leading LLMs, so researchers can query them through natural language interfaces they already know .

What Is the Gemini for Science Bundle?

Gemini for Science is not a single model. It is a suite of experimental AI tools announced by Google at I/O 2026 (May 19–20, 2026), designed to accelerate the scientific method . It is best understood as an "orchestration layer" for scientific research .

Gemini for Science bundles three core pillars:

• Hypothesis Generation — built on Co-Scientist, a team of AI agents that work together to generate, debate, and evaluate research hypotheses. Google published a same-day Nature paper validating the approach .
• Computational Discovery — built on AlphaEvolve and Empirical Research Assistance (ERA). ERA reportedly beat the CDC's forecasting model for infectious disease prediction in benchmarks published at launch .
• Literature Insights — built on NotebookLM, providing automated literature summarization and research analysis .

Additionally, Gemini for Science includes Science Skills, a separate layer in Google Antigravity that connects to over 30 life-science databases .

The experimental tools are available through Google Labs at labs.google/science, with gradual access beginning at launch .

Commercial Logic Behind the Partnership

For Google Cloud: SandboxAQ's LQMs represent a new, high-value workload category — scientific and quantitative AI — that drives cloud consumption. Google Cloud gains a differentiated, enterprise-grade product alongside its general-purpose Gemini models, positioning itself as the infrastructure provider for "AI-native scientific discovery" .

For SandboxAQ: The Google Cloud Marketplace provides immediate access to Google's vast enterprise customer base, a trusted billing and procurement channel, and scalable infrastructure — including integration with NVIDIA DGX Cloud on Google Cloud . SandboxAQ, an Alphabet spin-out now valued at $5 billion, deepens a natural commercial relationship .

What the Marketplace Listing Means for Scientific AI

This is a commercial milestone for scientific AI. It marks the first time specialized Large Quantitative Models — a category distinct from general-purpose LLMs — are available through a major cloud marketplace with enterprise procurement, billing, and security .

It signals that scientific AI is moving from research labs and custom deployments into a standardized, commercially listed product category that enterprises can buy off the shelf. SandboxAQ CEO Jack Hidary has said LQMs will eclipse LLMs in verticals like pharma, materials, energy, and finance , and the marketplace availability is the infrastructure that makes that shift real.

Key Takeaways

• LQMs are trained on physics, chemistry, and math — not text — and optimize for quantitative predictions rather than language generation.
• AQCat (materials) and AQPotency (drugs) are the first two LQM models available on Google Cloud Marketplace.
• Gemini for Science is an orchestration suite, not a single model, bundling Co-Scientist, AlphaEvolve, ERA, and NotebookLM.
• The partnership positions Google Cloud as the infrastructure backbone for scientific AI while giving SandboxAQ enterprise distribution.
• The marketplace listing represents the first time specialized LQMs are available through an enterprise cloud procurement channel.