Perplexity Secures Government-Wide Contract to Deliver Enterprise AI

Perplexity has signed a government-wide partnership with the U.S. General Services Administration (GSA) to make its Enterprise Pro for Government AI platform available to all federal agencies. The offering is accessible through the GSA Multiple Award Schedule (MAS IT) and will be provided at essentially no cost for up to 18 months. The company says the agreement includes long-term pricing commitments, a multi-model ecosystem and security assurances aimed at easing adoption inside agencies.

Detailed Blog Access: https://www.perplexity.ai/hub/blog/announcing-our-partnership-with-the-united-states-government

What the Agreement Unlocks for Agencies

Under the partnership, agencies can procure Enterprise Pro for Government directly through GSA’s MAS IT. Perplexity says the offering will be made available “at essentially no cost for up to 18 months,” along with negotiated long-term pricing meant to keep the service sustainable for agencies over time. The company frames MAS availability as a way to simplify procurement and give agencies access to an enterprise-grade AI platform without additional vendor layers.

Model Choice at the Center of the Platform

Perplexity highlights model flexibility as a primary feature of the partnership. According to the company, agencies have consistently asked for “choice”, and the Enterprise Pro for Government bundle is positioned as model-neutral. It includes Perplexity Sonar, GPT, Claude and Gemini. The architecture links these models to both an internet-scale knowledge index and an agency’s internal systems. The intention, as described by Perplexity, is to let administrators select the model best suited for a specific task and avoid reliance on any single provider.

Security and the FedRAMP Pathway

Security is a major focus. Perplexity says it is “the first (and currently only) publicly available AI platform to provide secure-by-default guarantees to all federal users, including users with no subscription or contract.” The company also points to its AI Prioritisation designation, described in the announcement as one of only two services with this status. According to Perplexity, this positioning accelerates its FedRAMP process through the cloud-native 20x pathway, which could shorten reviews and reduce administrative overhead for agencies seeking compliant AI tools.

Perplexity notes that tens of thousands of federal users rely on the platform for everyday tasks and “decisions of national importance”. The MAS availability is presented as a way to expand access across agencies. The company says use cases range from routine research, briefings and summarisation to high-stakes decisions where faster information retrieval can influence operations and policy.

Managing Vendor Risk in a Multi-Model World

Perplexity positions its multi-model design as a safeguard against vendor concentration. The company portrays the approach as a way to mitigate outages, model drift or compatibility issues from any single model provider and to maintain “uninterrupted access to frontier AI”. While this reduces reliance on one vendor in theory, agency teams will still need to manage integration complexity, version changes and governance practices when multiple models and data sources operate together.

Even with simplified procurement, federal agencies face several common challenges when adopting AI platforms. These include integration with legacy environments, governance of model behaviour and updates, and setting boundaries for sensitive or high-consequence decisions that require human oversight. Perplexity points to its security posture and GSA designation as support mechanisms, but each agency will have to evaluate the platform against mission-specific requirements before moving to broader deployment.

The GSA agreement removes a substantial procurement barrier, and the free access period could encourage early pilots and proofs of concept. For Perplexity, the MAS route creates a formal entry point into the federal market. For agencies, it promises faster access to multi-model AI capabilities while reinforcing the need for strong governance, auditing, and integration planning to ensure dependable results as usage scales.