DeepSeek API Pricing Change: Official Search and Fetch Endpoint Costs Begin October 31, 2026

Introduction: Verified API Pricing and Timeline

As of October 31, 2026, DeepSeek will implement new standalone pricing for its Search and Fetch API endpoints. Each successful Search or Fetch request will now be billed at $0.75 per 1,000 calls. This change is confirmed by DeepSeek’s official release notes and will take effect at 16:00 UTC on the specified date. Previously, these server-side tool calls were free for API users. The company has stated that failed, empty, or timed-out requests will not incur charges.

Sources: deepseekv4.network

This new per-request cost is a separate line item from token-based charges and applies to both Search and Fetch endpoints. The announcement has been corroborated by several independent trackers and industry reports, all confirming the specifics of the price, timing, and scope of the change.

Sources: infro.io · pricingsaas.com

Detailed Pricing Model and Billing Conditions

The fee for using the Search and Fetch endpoints has been set at $0.75 per 1,000 successful requests—defined as those returning a valid HTTP 200 status and non-empty results or page content. Calls that do not meet this criterion (such as errors or empty outputs) are not billed. This model offers some predictability for users but also requires them to distinguish between total calls and billable events.

Sources: deepseekv4.network · zentor.ai

Importantly, this new cost does not replace nor modify the existing token-based pricing; rather, users will now see two distinct charges on their billing statements: one for tokens and one for Search/Fetch tool calls. Existing rate limits and quotas reportedly remain unchanged, and no official information suggests otherwise.

Sources: infro.io · api-docs.deepseek.com

Workflow Example: Projecting Costs Under the New Model

For organizations or developers using DeepSeek’s APIs in automated agents or user-facing applications, estimating the new cost layer is now essential. A basic forecast formula, as outlined by industry commentators, is: cost = (number of expected successful Search/Fetch calls) × $0.00075. For example, if a workflow generates 10,000 successful Search calls per month, the additional monthly cost would be $7.50, exclusive of token charges.

Sources: infro.io

Since only successful calls are billed, users are encouraged to monitor both total call attempts and the 'success rate' of those calls. This will allow for accurate budgeting and prevent unnecessary expenditures due to over-calling or poorly optimized retrieval strategies.

Sources: infro.io

Market Comparison: DeepSeek’s Pricing in Context

DeepSeek’s newly introduced tool-call pricing is competitive when compared to other major AI platform providers. According to industry analysis, other providers may charge between $10 and $14 per 1,000 similar search calls, making DeepSeek’s $0.75 per 1,000 calls a much lower-cost alternative. This makes the platform attractive, particularly for higher-volume search applications where cost per call is a critical factor.

Sources: infro.io

Still, the zero-cost period for Search and Fetch ends with this change, so overall operating expenses are likely to increase for existing DeepSeek API users—even if the incremental per-call cost is modest relative to competitors.

Sources: infro.io

Current Unknowns and Regional Pricing Ambiguity

It remains unclear whether the '5 yuan per 1,000 calls' figure, reported by some third-party sources, represents official Chinese-market pricing or is merely a conversion for illustrative purposes. No official DeepSeek announcement has clarified regional or currency-specific pricing policies as of this update. Similarly, there is no confirmed information about promotions, credits, or large-volume discounts for these endpoints.

Sources: www.tokenpost.com · zentor.ai

No modifications to the existing rate limits or quotas have been announced or documented. Users should rely on the official API documentation to track any changes, as industry sources state that such parameters remain the same.

Sources: api-docs.deepseek.com

Practical Steps and Recommendations for API Users

In anticipation of the billing change, users should immediately begin tracking their actual Search and Fetch usage and monitor success rates. This information will be crucial for accurate budgeting and preventing unexpected overages when the new pricing takes effect.

Sources: infro.io

Organizations with higher volumes or those concerned about cost escalation should consider contacting DeepSeek for enterprise agreements or volume discounts, even though none are currently published. Updating financial models to reflect tool-call costs and analyzing alternative strategies—such as caching or constrained querying—may help optimize spend.

Sources: infro.io

Research limitations

Third-party reporting only: an official announcement was not established in this research. Verify before publishing. No changes to rate limits or usage quotas are stated; the new price per unit is confirmed.

Researched with AI. Sources and claims require editorial review.