This page is dedicated to resources for government personnel: Federal, state, and local!

7 Steps to AI Readiness

Here’s a framework to prepare your government organization for AI.

Fill this out for your own government organization with this worksheet:

Gov Org AI Readiness Worksheet

Metrics for Success:

1. Guarantee baseline AI access for every employee

Intent: Improve AI literacy across the workforce and begin measuring where AI delivers productivity gains.

AI Chatbot Products Authorization
Google Gemini FedRAMP High, IL-5
Microsoft Copilot IL-5 (Sept. 2025)
ChatGPT Government None
ChatGPT Enterprise None
Anthropic Claude None
DHSChat CUI/IL-5
NIPRGPT IL-5
CamoGPT IL-5+6
AskSage FedRAMP High, IL-5+6, TS

2. Subsidize & incentivize compute/GPUs

Intent: Ensure your organization can run AI models efficiently within authorized environments.

DoD HPC Compute (unclass)

DoD HPCMP Location A100 V100 T4 A40 L40 MI250X MI300A (APU) Total GPUs
Navy DSRC (Stennis) 128 112 0 16 24 0 128 408
AFRL DSRC (Wright‑Patt) 152 104 0 30 0 0 0 286
ARL DSRC (Aberdeen) 264 136 512 24 0 64 0 1,000

Cloud Compute

Cloud GPU Types Publicly Confirmed
AWS GovCloud (US) (IL-5) A100 (P4d), G4dn/T4, G5/A10G
AWS Secret Region (IL-6) H100 (P5)
Azure Government (IL-5) ND A100 v4 / ND H100
Google (IL-5) A3 (H100), G2 (L4) (Vertex/Compute)
Oracle (IL-5) H100/H200/Blackwell families
Model Service Cloud Authorization(s) Available Models
OpenAI Azure IL-5+6, TS OpenAI GPT-4o
AzureML Azure IL-5 Meta, Mistral
Bedrock AWS IL-5, TS Anthropic, Meta, Mistral
Vertex Google IL-5+6 Google, Anthropic, Meta, Mistral

3. Publish clear, stable AI standards

Intent: Provide vendors and internal teams with a clear performance and safety baseline.

Proposed Agent Access

Here’s a DoD Benchmark: JointStaffBench, created by GovBench.

Here’s a DoD Benchmark: JointStaffBench, created by GovBench.