Harvey Reports 14 Times Increase in Token Usage in Six Months
Harvey’s co founder and President, Gabe Pereyra, has reported a significant increase in the number of tokens processed on the legal AI platform, with usage rising fourteen times over the past six months. This surge indicates a growing reliance on AI technology within the legal sector.
The reported usage data does not account for embedding tokens, nor does it disclose the associated costs of this growth. It is understood that Harvey is currently developing a new pricing model for its customers to accommodate these changes.
In a recent discussion on X, Pereyra suggested that it may become increasingly common for law firms to pass on token costs to clients. He noted that law firms already transfer costs related to other legal technologies, making this a plausible development.
This announcement comes at a time when competitor Legora has also indicated a shift to a pricing structure that considers token consumption. The trend reflects a broader increase in AI usage across various sectors, with clients increasingly demanding that law firms reduce their fees in light of AI utilisation.
In addition to Harvey, other platforms such as Claude for Legal, Codex for Legal, and Perplexity are expanding their engagement in legal technology. Both law firms and legal tech companies are exploring open source solutions, which could potentially lower costs.
A key challenge in this evolving landscape is that not all tokens have the same cost structure. For instance, employing advanced models for complex legal issues not only consumes more tokens but also incurs higher expenses. The use of agentic tools can further escalate costs, as these tools may require multiple iterations to achieve optimal results.
Legal professionals often prioritise minimising risk, which complicates decisions regarding cost effective AI solutions. When faced with options that vary in price and accuracy, lawyers are likely to favour the more expensive, higher performing tools, particularly when accuracy is paramount in legal contexts.
Legal AI platforms may possess a competitive edge over traditional model providers by allowing users to switch between various models, potentially managing costs more effectively. However, reliance on a single model could lead to substantial expenses, raising questions about the feasibility of frequent model switching.
The issue of cost allocation is critical. As tokens become an increasingly prevalent currency within the legal industry, law firms must consider whether to pass these costs onto clients. It is argued that clients ultimately bear the financial burden of law firm expenses, regardless of how these costs are structured.
The implications of rising AI costs may prompt corporations to internalise more legal work to maintain control over expenses. Conversely, external legal providers may engage in price competition to retain business, potentially triggering a price war among legal AI companies.
As the legal sector adapts to these changes, the dynamics of cost management and service delivery are shifting. Clients are demanding reductions in fees while the costs of AI tools continue to rise, creating a complex and unsustainable situation.
The current market is characterised by uncertainty, with the potential for significant changes in the legal business model. As the industry navigates this transitional phase, it is likely that some firms will emerge as leaders while others may struggle to adapt.