GSA offers federal agencies 40% discount on Anthropic Claude models

The expanded OneGov deal

The General Services Administration (GSA) has officially expanded its OneGov agreement with Anthropic. This deal provides federal agencies with access to Anthropic's Claude models under a new set of terms. The primary goal of the expansion is to provide government entities with more flexible and cost-effective ways to integrate artificial intelligence into their workflows.

Under the new terms, federal agencies will receive a 40% discount on token-based usage. A token is a basic unit of text that AI models use to process and generate information. By offering this discount, the GSA aims to make high-performing models more accessible to a wider range of government departments. The agreement is designed to help agencies move away from a heavy dependency on temporary promotional pricing.

This move comes as the original agreement reached its extended deadline. The GSA stated that it will continue to renew agreements where appropriate to ensure continued access to critical tools. The agency's stated priority is to deliver the best value for taxpayers while maintaining the availability of these technologies. As reported by FedScoop, this expansion provides a clearer path for agencies looking to adopt Claude models consistently.

Moving to consumption-based pricing

The new agreement marks a significant shift toward consumption-based pricing for the federal government. In this model, agencies pay for what they actually use rather than paying a flat fee per user. This mirrors a broader trend seen in the private sector, where consumption-based models have rapidly replaced traditional seat-based pricing. The GSA noted that this specific pricing model is designed to scale with actual use by the government.

Anthropic's deal follows a similar path taken by OpenAI, which recently switched to consumption-based pricing as part of its own OneGov deal. By removing minimum purchase requirements, the GSA allows agencies to start small and scale up as their needs grow. This flexibility is intended to accommodate different agency sizes and varying levels of AI integration across different departments.

Before this expansion, Anthropic had offered a $1 per user price point for Claude for Government. That offer was part of a temporary extension that was set to end in October. The transition to the current model represents a more permanent shift in how the government procures large-scale AI services. It allows for a more granular approach to spending where the cost is directly tied to the volume of data processed.

Predictability and budgeting hurdles

While the new pricing model offers flexibility, it also introduces significant challenges for government budget offices. Analysts have expressed concern that consumption-based pricing makes it difficult to predict total costs accurately. Because vendors often lack transparency into exactly how token consumption is calculated and billed, agencies may find it hard to plan their annual budgets. This lack of clarity can create friction during the procurement and planning phases.

The concern is that a small group of power users could drive a disproportionate share of the total costs. If a few employees use the models very heavily, it could turn a successful productivity win into a significant budgeting challenge. This dynamic makes it harder for administrators to ensure that spending remains within the limits set by oversight bodies. Agencies must now balance the benefits of high-volume usage with the need for fiscal control.

Industry experts have highlighted these risks as agencies begin to adopt these tools more broadly. The difficulty in forecasting costs is a known issue for many organizations moving away from fixed-fee software. For the federal government, where budget transparency is paramount, these unpredictable costs could lead to complications. Agencies will need to develop new internal monitoring systems to track token usage in real-time.

The broader federal AI landscape

The GSA's actions are part of a larger effort to standardize how the government accesses various AI technologies. While Anthropic's relationship with the federal government remains solid in most areas, the company is currently involved in an ongoing legal battle with the Pentagon. Despite this specific legal hurdle, the OneGov framework continues to serve as a primary vehicle for distributing AI tools to various agencies.

Other major players are also participating in the OneGov framework with varying terms. For example, Google is currently offering access to its Gemini models for $0.47 per agency. However, that specific offer is set to expire on November 15. The GSA has indicated that it will review these offers and renew agreements where it sees the most value for the public.

The competition between these providers shapes how federal agencies choose their primary AI partners. By offering steep discounts and flexible pricing, Anthropic is positioning itself as a viable alternative to other major models. The GSA's role is to navigate these various offers to ensure that the government maintains access to the most effective tools. This ongoing process highlights the complexity of managing a diverse portfolio of AI technologies across a massive bureaucracy.