Cheaper tokens met agentic workflows and legal tech spend still rose. Without model routing and usage caps, the bill grows in the dark.
Via Law.com: The Token Cost Illusion: Why Falling AI Prices Will Not Save Your Legal Departments Budget
If your legal AI budget assumes falling per-token prices will keep the line item quiet, Law.com has bad news. Bobby Balachandran reports that per-token prices across major AI providers fell roughly 75 percent year-over-year - from around USD 10 to USD 2.50 per million tokens in Ramp enterprise spending data - while legal technology spending grew nearly 40 percent above pre-GenAI baseline levels by the end of 2025, citing the Thomson Reuters / Georgetown Law 2026 State of the US Legal Market report.
That is the token cost illusion for owner-led firms and in-house teams alike: unit prices drop, agentic workflows multiply tokens, and the people clicking Generate are often not the people who must explain the invoice.
Law.com notes legal work is already document-heavy. A contract review with a full document, detailed prompt, and structured output can require 10,000 to 100,000 tokens in a single pass. Multi-jurisdictional monitoring, M and A diligence, and e-discovery can consume 50 to 500 times more tokens than a simple chatbot query. Agentic workflows average 1 to 3.5 million tokens per task when retries and self-correction are included.
Shadow AI makes the fog thicker. Personal ChatGPT subscriptions, browser tools, and departmental accounts outside IT mean you cannot forecast what you cannot see.
Model routing is the quiet lever. Law.com says a single routing decision can swing monthly costs by a factor of five at production volume with no obvious change in output quality. An illustrative NDA review on a frontier reasoning model can cost roughly 50 to 200 times more per document than the same workflow on a lighter model that may still hit the bar for structured tasks when properly evaluated.
On OpenRouter, agentic token consumption rose from 11 percent to over 50 percent of all usage by late 2025. Legal document review and discovery look a lot like the coding workflows that drove that curve. If you still budget from chatbot pilots, production agents will rewrite your math by an order of magnitude.
Practical budgeting guidance in the piece: multiply estimated token usage by 1.7 to 2.0x for retries, system prompts, and context overhead. Most pilot estimates skip that multiplier, which is why production reliably overruns the approved amount.
Privilege and confidentiality ride along. Every document sent to an external commercial API adds third-party processing risk. Private or on-prem options are not just IT shopping; they are governance decisions with cost and confidentiality payoffs.
Enterprise AI agreements running 12 months or less, as Law.com notes, also change negotiation leverage. Portable, model-agnostic architectures help buyers push back when vendors reprice. Workflows locked to one provider absorb whatever the market delivers.
Treat AI like any high variable cost: visibility, tiering, and ownership.
Per-token prices fell roughly 75 percent year-over-year, yet legal technology spending grew nearly 40 percent above pre-GenAI baselines by end of 2025.
AgentsROI does not sell you a single model. We help owner-led professional firms see what AI is actually costing and where it pays.
A Workflow ROI Audit maps chatbot pilots versus document-heavy agents, measures token and human-review cost per workflow, and flags where routing to a lighter model would cut spend without cutting quality. Model Selection and Continuity Planning turns tiering into a written policy with fallbacks when a vendor reprices. If personal accounts are already in the wild, a Shadow-AI Risk Assessment brings usage into view before finance finds it first.
Law.com's thesis is operational: workflow design, model routing, and governance decide cost more than the sticker price of a million tokens. Capture the value without surrendering the budget.
If your legal or professional services team needs a vendor-neutral read on agentic unit economics, book a Workflow ROI Audit with AgentsROI. We run the AI. You run the business.
This article summarizes publicly reported information and is for general informational purposes only. It does not constitute legal, tax, financial, investment, security, or compliance advice. AgentsROI.ai is not a law firm, accounting firm, or registered investment adviser. Facts, pricing, statistics, and product capabilities cited here reflect the sources listed at the time of writing and may change. Readers should verify current information independently and consult qualified professionals regarding obligations specific to their industry, jurisdiction, and circumstances - including applicable New York State and New York City requirements. AgentsROI.ai may have commercial relationships with vendors mentioned; where material, such relationships are disclosed. Nothing in this article is an endorsement of any specific AI product, model, or provider.
