GSA offers federal agencies 40% discount on Claude AI usage

A shift to consumption-based pricing

The General Services Administration (GSA) has announced a new pricing structure for Anthropic's Claude AI models. Under the OneGov program, federal, state, and local agencies can now access a 40% discount on token-based usage. This marks a significant transition in how the government procures and pays for large-scale artificial intelligence services.

A token is the basic unit of measurement for how much text an AI model processes or generates. By moving to a token-based model, the government can pay for the specific amount of data handled by the AI rather than paying a flat fee. This approach is designed to align costs more closely with the actual volume of work performed by the software.

The move follows a previous agreement where federal agencies had access to Claude for Enterprise and Government for $1 per user. That specific arrangement is set to expire on October 31. The new deal replaces that fixed-user model with a more fluid system that allows for variable usage levels across different departments.

How the OneGov program works

The OneGov program is a framework designed to streamline how government entities purchase technology and services. By using this program, agencies can bypass some of the traditional hurdles of individual procurement. The GSA aims to provide these entities with better pricing and more standardized access to modern tools like generative AI.

Under the new Anthropic agreement, agencies are not required to meet any minimum purchase requirements. This flexibility means a small department can use the same discounted rates as a large federal agency. The pricing is intended to be accessible through major cloud platforms and various authorized resellers, making it easier for different types of government bodies to adopt the technology.

This model mirrors a similar deal brokered by OpenAI with the GSA earlier this month. That agreement provided a 50% discount on token-based usage through a 27-month-long contract. The GSA is positioning these consumption-based models as the logical next step for integrating AI into regular government operations.

Financial impact and savings

Anthropic estimates that the price cuts provided through the OneGov agreement will save the government $1 billion. These savings are a primary goal of the GSA's efforts to manage the costs of rapidly expanding technology budgets. By leveraging the OneGov framework, the government can achieve economies of scale that individual agencies might not reach on their own.

The GSA reported that the OneGov program has already generated more than $1.8 billion in total savings. Of that total, roughly $1.6 billion was achieved specifically through various AI agreements. These figures highlight the scale at which the government is currently investing in and optimizing its use of artificial intelligence technologies.

The transition to discounted token-based pricing is intended to make it more financially viable for agencies to experiment with AI. Instead of committing to large, potentially unused licenses, departments can scale their usage up or down based on real-time needs. This helps prevent waste while ensuring that high-demand departments have the resources they need.

Current agency adoption trends

The move to a more flexible pricing model comes as federal agencies show increased interest in adopting Claude for their specific workflows. Just last week, the Department of Health and Human Services announced its intent to award a sole-source contract for Claude models. This indicates a growing trend of specific departments choosing Anthropic's models for their unique operational requirements.

The shift toward consumption-based access is a response to the fact that AI is being integrated into more regular operations. As agencies move away from occasional use cases toward daily workflows, the need for a scalable pricing model becomes more apparent. The GSA is trying to keep pace with this rapid integration by offering more adaptable procurement options.

Agencies will now have the ability to access these discounts directly through established cloud providers. This means the infrastructure for using Claude is already being built into the cloud environments that many government offices already use. It simplifies the technical and administrative work required to get the AI into the hands of government employees.

What comes next for government AI

The new agreement with Anthropic officially went into effect on Tuesday. This puts the government in a position to quickly deploy Claude models with a clear understanding of the costs involved. It also sets a precedent for how other AI providers might structure their deals with the federal government in the future.

As the OneGov program continues to expand, we can expect more agencies to move toward these types of consumption-based contracts. This trend likely reflects a broader shift in government procurement toward software-as-a-service models. It allows for a more dynamic relationship between the government and technology providers, where costs are tied to actual output.

The success of these deals will likely be measured by how effectively they help agencies integrate AI into their daily tasks. With the GSA focusing on both cost-saving and accessibility, the next few months will be critical for seeing how these models are actually used. The goal remains to make the government smarter and more efficient by providing the right tools at a manageable price.