WHY THIS MATTERS
DeepSeek announced this week a price increase of up to 1,100% for its V4 API models, with rates now varying by peak versus off peak hours, token type, and cache hit or miss. V4 Pro output tokens, for example, jump from $0.87 to $3.96 per million at peak. The company says it wants to allocate resources more reasonably, but the move confirms the warning it gave on August 12, when V4 Pro exited preview. For Brazilian agencies that relied on DeepSeek as a cheap option, the change requires reassessing costs and avoiding dependence on a single vendor.In this article
DeepSeek Announces Major API Price Increase
On August 13, 2026, DeepSeek announced a steep increase in API pricing for its V4 models, with the new rates taking effect this week, between August 16 and 17. According to the company, the increases range from 50% to 1,100%, depending on the model, the token type (input or output, cache hit or miss), and the time of day, since DeepSeek is introducing a tiered structure that distinguishes peak from off peak hours. The announcement marks a significant shift for a company that had built its reputation on offering advanced models at prices well below the market average.
The Numbers: Peak, Off Peak, Input, Output, Cache
The examples released show the scale of the change. V4 Pro output tokens will now cost $3.96 per million tokens at peak hours, more than four times the prior rate of $0.87 per million; during off peak hours, the rate is cut in half, to $1.98 per million. V4 Flash cache miss input tokens rise to $0.44 per million at peak, up from $0.14 per million. V4 Pro cache miss input tokens climb from $0.435 to $1.32 per million at peak hours. Every combination of model, token type, and time slot now carries its own rate, which demands closer cost monitoring from anyone running production workloads.
DeepSeek's Stated Reasoning
DeepSeek says the goal is to allocate resources more reasonably. The new tiered structure, with higher rates at peak hours and lower ones off peak, is meant to encourage developers to shift non urgent workloads toward periods of lower demand on the company's infrastructure. In effect, DeepSeek is using price as a capacity management tool, a common practice among cloud providers but one the company had not previously applied to its own pricing.
How It Compares to Competitors
Even after the jump, DeepSeek's new peak rates remain below some competitors. For comparison, Anthropic's Fable 5 model charges $50 per million output tokens, more than twelve times the new peak ceiling for V4 Pro. DeepSeek is moving from being an ultra cheap option to a still competitive one, a shift that changes how it should be positioned in any cost comparison between providers.
The Warning Given on August 12
The increase confirms a signal DeepSeek had already sent. On August 12, when it moved V4 Pro out of preview into general availability (build 0813), the company's pricing page already noted that a significant increase was coming, without specifying a date or magnitude. A week later, the numbers show that warning was not just rhetoric: the increase arrived quickly and in proportions that likely exceeded what many developers expected.
Why This Matters for Brazilian Agencies and SMBs
For marketing and customer service automation agencies in Brazil, DeepSeek had been one of the cheapest options among advanced models for building high volume chatbots and agents. A jump of four to twelve times on certain token types changes the cost math overnight for anyone who anchored pricing or margins to DeepSeek's old rates. The moment calls for immediately retesting cost per interaction, considering scheduling batch workloads for off peak hours, and above all caution against locking long term client contracts to any single vendor's promotional pricing. Diversifying providers and building repricing clauses into contracts is no longer a theoretical precaution, it is now a practical necessity.