Pricing

A private research lab, with capacity you can plan around.

Every paid plan includes a dedicated research environment, a hard usage ceiling, and a clear estimate of GPT-5.6 Luna research capacity.

Core

$0 forever

The Apache-2.0 engine and typed SDK for teams that operate their own stack.

Included research capacityBring your own modelsCapacity depends on your infrastructure

Private computeSelf-hosted

Local CLI and API
Protocol and approval gates
Evidence ledger and exports
Community support

Use the engine

Recommended

Cloud Team

$2,500 per month

A private research workspace for focused teams running one investigation at a time.

Included research capacity≈125 Luna research hoursAbout 1–2 paper-scale research packages

Private computeDedicated 4 vCPU · 16 GB RAM · 100 GB

Up to 5 members
One active investigation
Private repositories and data
Hard budgets and approvals

Start a conversation

Cloud Program

$7,500 per month

A governed portfolio for concurrent R&D programs, connectors, and expert review.

Included research capacity≈375 Luna research hoursAbout 3–5 paper-scale research packages

Private computeDedicated 8 vCPU · 32 GB RAM · 200 GB

Multiple active investigations
Program budgets and milestones
Expanded compute and connectors
Expert scientific QA

Start a conversation

Enterprise

$125k+ per year

Dedicated or customer-controlled deployment for enterprise and government teams.

Included research capacity≈6,250 Luna research hours / yearAbout 40–60 paper-scale research packages

Private computeDedicated 16 vCPU · 64 GB RAM · 500 GB

Dedicated or BYOC deployment
SSO, audit and SLA
Residency and procurement
Private implementation support

Start a conversation

Capacity estimates, not publication promises.

Research hours use GPT-5.6 Luna at a planning rate of approximately two million blended tokens per autonomous research hour. A paper-scale package assumes roughly 100–150 research hours and can include a manuscript, figures, provenance, code and replication materials. Actual consumption varies by domain, tools, source volume and review depth.

Model choice

Spend capacity on depth or volume.

Choose Luna for the greatest investigation volume, Terra for balanced depth and cost, or Sol for the hardest reasoning. ApolloBot previews the impact on remaining research hours before a run begins.

LunaMaximum hours

Terra≈40%

Sol≈20%

Design partner

$15,000

60 days and one scoped investigation; eligible for annual credit.

Managed sprint

$20–40k

Four to eight weeks, evidence through replication package.

Phase I program

$50–100k

Feasibility campaign, risk gates and Phase II blueprint.