Skip to main content

​
General

The SandboxAQ MCP Server exposes SandboxAQ’s proprietary scientific AI models as callable tools inside Claude and other MCP-compatible clients, so you can run high-accuracy scientific calculations conversationally. See the SandboxAQ MCP Server overview for the full description.
Claude Desktop, claude.ai, Claude Science, Claude Code, and Antigravity all have documented setup flows — the Quickstart has condensed steps for each, and Connect your MCP client has the full walkthrough. Config details for Antigravity are on the SandboxAQ MCP Server page. Any other client that speaks the Model Context Protocol over Streamable HTTP can connect using the same server URL and OAuth flow.
No. Authentication is OAuth-based. You sign in with your SandboxAQ email and password or with Google. No API keys are issued or accepted.
AQPotency predicts how strongly a compound binds a protein target, and AQCat predicts how molecules adsorb onto catalyst surfaces, so you can rank compounds, screen libraries, and evaluate catalyst candidates without running the underlying simulations yourself. See the AQPotency and AQCat articles for full descriptions.
AQPotency and AQCat address different scientific domains: AQPotency scores protein-ligand pairs for drug discovery, while AQCat computes adsorption energies on metal surfaces for catalysis and materials work. They are billed separately but share the same connection, login, and server URL. See the Pricing page for how each is metered.
No coding is required. You describe the result you want in plain language and your MCP client selects the appropriate tool and fills in the parameters for you, so the entire workflow runs conversationally. See Connect your MCP client for worked examples.
No local compute is required. All inference runs on SandboxAQ-managed infrastructure, so you need only an MCP-compatible client and an internet connection to run calculations. See the SandboxAQ MCP Server overview for how requests are executed.

​
Access and setup

Self-serve: register at platform.aisim.sandboxaq.com/signin with an email address and password or with Google, accept the subscription terms, and add a payment method. Your account is active immediately and your server URL is https://mcp.platform.aisim.sandboxaq.com. The Quickstart walks through it.Enterprise accounts are provisioned separately by the SandboxAQ team and are issued their own host — contact mcp-prod@sandboxaq.com. See Contact for other enquiries.
In the platform console under Account → MCP, or via the Copy MCP URL button in the top-right corner. Self-serve accounts use https://mcp.platform.aisim.sandboxaq.com; enterprise accounts are issued their own host.
See Troubleshooting for more.

​
Enterprise plan

Enterprise is a negotiated annual plan for production R&D, automated pipelines, and LLM-orchestrated workflows. It adds higher rate limits, structured support with defined response targets, priority model upgrades, and deployment options beyond shared managed access. The full Individual vs. Enterprise comparison is on the Pricing page.
Contact mcp-prod@sandboxaq.com. Enterprise accounts are provisioned by the SandboxAQ team rather than through self-serve sign-up, and the terms — pricing, SLAs, deployment model, and volume commitments — are negotiated separately from the Individual tier terms published on this site.
Yes. Self-serve accounts all share one host, but each enterprise account is issued its own dedicated host, so your MCP URL will not match the one shown throughout this site. Enterprise URLs take the form https://mcp.{tenant}.aisim.sandboxaq.com, where {tenant} identifies your organization. Use the exact URL shown in your platform console under Account → MCP.
Dedicated onboarding, a technical account manager, and structured support with defined severity classes and first-response targets. See Enterprise Tier Support for support hours and the full severity table. Those terms apply to the Enterprise plan only — Individual-tier support is best effort through support.aisim@sandboxaq.com.
Enterprise customers can reach out about Cloud Marketplace containers on AWS, Azure, and GCP, which run in their own tenant and cloud hardware. Contact mcp-prod@sandboxaq.com to scope it.

​
Trust and data

AQPotency is calibrated for ranking compounds rather than reporting absolute affinities, so predictions should be used to order candidates rather than treated as measured values. Every prediction returns a potency_sigma uncertainty estimate alongside ligand_similarity and protein_in_training applicability-domain signals at no additional cost. See Interpreting Results for guidance on reading them.
Each organization’s requests and usage data are isolated on the platform, so cross-organization access is not possible. See the Privacy Policy and Data Protection Addendum for full details on how data is collected, used, and retained.

​
Pricing and billing

New accounts include $2,000 in free credits for the first 30 days — about 2,000 AQCat runs or 2 million AQPotency protein–ligand pairs — so you can evaluate both models before committing. Billing starts once your credits run out. See Pricing for per-call costs.
AQCat is billed at $1.00 per Relaxation Unit. AQPotency is tiered: $0.01 per pair for the first 10 pairs of a call, then $0.001 per pair after that. See Pricing for worked examples.
Not on the Individual tier. Enterprise is $50,000 / year per org or department and includes higher rate limits, an SLA, and dedicated support. See the tier comparison.
Your balance is exhausted. Add or update your payment method in the platform console and retry.

​
Support

Individual tier: email support.aisim@sandboxaq.com or use the contact form. Enterprise plan: your technical account manager, or mcp-prod@sandboxaq.com for contracts, SLAs, and custom deployments.
On the Individual tier, support is best effort — we respond as quickly as we can, but no response time is committed. Committed support hours, severity classes, and first-response targets apply to the Enterprise plan only; see Enterprise Tier Support.