DeepSeek API Pricing and Cost per Workflow
By Paige Gilmore, Founder, NetLift · Published 2026-07-28 · Updated 2026-08-11
Prices checked 2026-08-16 against official vendor pricing pages.
DeepSeek API pricing ranges from $0.30 to $1.05 per million input tokens depending on the model and priority tier. The v3.2 US-based model charges $0.62 for input and $1.85 for output per million tokens.
DeepSeek offers a tiered pricing structure for its v3.1 and v3.2 models, primarily differentiated by regional availability and processing priority. For v3.1, costs are determined by whether a workflow utilizes Flex, Standard, or Priority input tiers, allowing teams to trade off latency for lower operating expenses.
Engineering and FinOps teams should note the significant delta between input and output costs in the v3.2 US-based deployment. While input remains relatively inexpensive, output tokens are priced higher, making the verbosity of model responses a primary driver of total workflow cost.
Official per-token prices
| Model / rate | Price | Unit |
|---|---|---|
| DeepSeek v3.1 — Flex Tier Input (Sydney) | $0.3 | per 1M input tokens |
| DeepSeek v3.1 — Priority Tier Input (Sydney) | $1.05 | per 1M input tokens |
| DeepSeek v3.2 — On-Demand Standard Input (US) | $0.62 | per 1M input tokens |
| DeepSeek v3.2 — On-Demand Standard Output (US) | $1.85 | per 1M output tokens |
How do the Sydney priority tiers affect budgeting?
The Sydney-based v3.1 model provides three distinct price points for input tokens. The Flex tier offers the lowest possible cost, while the Priority tier increases the rate to handle urgent workloads. Standard input sits between these two, providing a middle ground for consistent daily operations.
What are the cost considerations for DeepSeek v3.2 in the US?
For the v3.2 model deployed in the US, pricing follows a standard on-demand structure. Unlike the Sydney v3.1 tiers which focus on input, the v3.2 model explicitly separates input and output costs. This requires a different forecasting model that accounts for the specific token ratio of your application's prompts and completions.
Determining whether DeepSeek spend is effective requires measuring the specific time saved by developers or end-users compared to the baseline cost of these API calls. NetLift provides the framework to correlate these infrastructure expenses with business outcomes, ensuring that shifting from a Flex to a Priority tier actually results in a measurable increase in net return.
Sources
All pricing on this page comes from official vendor pages:
- https://aws.amazon.com/bedrock/pricing/ — retrieved 2026-08-11
Frequently asked questions
What is the lowest input cost available for DeepSeek?
The lowest rate is $0.3 per 1M input tokens, available through the DeepSeek v3.1 Sydney Flex tier.
How much does DeepSeek v3.2 cost for US-based workflows?
The v3.2 model in the US costs $0.62 per 1M input tokens and $1.85 per 1M output tokens.
What is the difference between v3.1 Sydney Standard and Priority pricing?
Standard input is priced at $0.6 per 1M tokens, while Priority input is $1.05 per 1M tokens.
About the author
Paige Gilmore is the founder of NetLift, the AI Value Management platform that helps organisations measure the cost, savings and return of AI adoption. Paige Gilmore on LinkedIn