Tuesday, August 18, 2026
DeepSeek has raised prices for its AI API and introduced scheduled peak and off-peak rates, making time of day a new factor in what businesses pay for AI usage.
DeepSeek, a Chinese AI company, has changed how it charges businesses to use its models.
New pricing that took effect August 16 raises the company’s API rates and introduces different prices depending on when customers use the service.
Think of it as scheduled surge pricing for AI.
DeepSeek now charges twice as much during seven peak hours each day as it does during off-peak periods. Unlike traditional surge pricing, those higher-priced hours are set in advance.
Prices Went Up Too
DeepSeek also raised its underlying usage prices.
For its higher-priced model, output previously cost $0.87 per million tokens. Under the new pricing:
Its lower-priced model increased from $0.28 per million output tokens to $0.66 off-peak and $1.32 during peak hours.
DeepSeek also charges for information sent into its models, and those rates now follow the same peak and off-peak schedule.
The bigger change for businesses is that AI usage now costs more, and timing can affect the bill.
Seven Hours a Day Carry the Higher Price
DeepSeek’s peak periods run from 01:00 to 04:00 UTC and 06:00 to 10:00 UTC each day. All other hours are off-peak.
The company says the pricing is intended to spread demand more evenly and encourage customers to schedule work during lower-use periods.
Some AI work can move. Internal processing or analysis that doesn’t need an immediate answer could potentially run during cheaper hours.
Customer-facing uses have less flexibility. AI supporting customer service, product experiences or other real-time interactions generally has to run when customers need it.
So the impact of the new pricing will depend heavily on how a business uses AI.
INSIDER TAKE
DeepSeek’s pricing change shows how AI costs are starting to behave more like other variable operating expenses.
As companies put AI into customer service, products and internal operations, usage charges become part of the ongoing cost of running the business.
DeepSeek has now added time of day to that cost.
The model is closer to scheduled surge pricing or airline pricing than the fixed software pricing many executives are used to. The underlying service may be the same, but the price changes depending on when it is used.
The increase itself is significant. For DeepSeek’s higher-priced model, output went from $0.87 per million tokens to $1.98 during off-peak periods and $3.96 during peak hours.
Some businesses may be able to move non-urgent AI work to cheaper periods. Uses tied directly to customers may have far less flexibility.
As AI becomes a bigger part of subscription operations, pricing changes like this make the cost of using it worth watching alongside what the technology can do.
Related Member Resource
AI pricing changes are another reminder that technology costs can shift after a tool becomes part of everyday operations. Members evaluating or replacing technology can use our workbook to define requirements, compare vendors, and examine costs and tradeoffs before committing.
Sources
About Subscription Insider®
Subscription Insider helps executives make better decisions across the business of subscriptions. Decisions involving growth, pricing, retention, billing, compliance, and customer experience are connected, but they rarely sit within one department. Since 2009, Subscription Insider has combined independent reporting and analysis, real operating experience, and practical guidance to help leaders see what’s changing, understand what it means for their businesses, and decide what to do next. Subscription Insider serves executives leading subscription, membership, and recurring revenue businesses. Learn more at SubscriptionInsider.com.