DeepSeek API pricing starts at $0.3 per 1M input tokens for v3.1 Sydney (Flex) and $0.62 per 1M input tokens for v3.2 in US regions. Output tokens for v3.2 US deployments are priced at $1.85 per 1M.
14-day free trial. No credit card required.
DeepSeek API costs are determined by model version, geographic region, and request priority. For v3.2 workflows in US regions, the price is $0.62 per 1M input tokens and $1.85 per 1M output tokens.
In the Sydney region, v3.1 pricing is segmented into three tiers to manage latency and cost. The Flex tier is the most affordable at $0.3 per 1M input tokens, while the Priority tier is $1.05 per 1M input tokens.
| Model / rate | Price | Unit |
|---|---|---|
| DeepSeek v3.1 — Flex input tokens (Sydney) | $0.3 | per 1M input tokens |
| DeepSeek v3.1 — Standard input tokens (Sydney) | $0.6 | per 1M input tokens |
| DeepSeek v3.1 — Priority input tokens (Sydney) | $1.05 | per 1M input tokens |
| DeepSeek v3.2 — Standard input tokens (US Regions) | $0.62 | per 1M input tokens |
| DeepSeek v3.2 — Standard output tokens (US Regions) | $1.85 | per 1M output tokens |
14-day free trial. No credit card required.
For teams using v3.1 in Sydney, the choice between Flex, Standard, and Priority tiers significantly impacts the budget. While the Flex tier costs $0.3 per 1M input tokens, moving to Priority more than triples that cost to $1.05. Engineering leaders must determine if the performance requirements of their specific workflow justify the higher per-token rate.
Workflows running v3.2 in US regions follow a standard input/output pricing model. At $1.85 per 1M output tokens, the cost of generation is approximately three times higher than the $0.62 input rate. FinOps teams should use these figures to model total cost of ownership for high-output applications like code generation or document summarization.
Calculating the ROI of DeepSeek requires comparing total API spend against the baseline cost of manual labor. NetLift allows engineering teams to measure the time saved per workflow and compare it to the $0.62 per 1M input rate, ensuring that the shift to automated intelligence produces a positive net return for the business.
14-day free trial. No credit card required.
All pricing on this page comes from official vendor pages:
14-day free trial. No credit card required.
About the author
Paige Gilmore · Founder, NetLift
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 LinkedInInput caching reduces API expenses by allowing providers to reuse previously processed context. For gpt-5.6-sol, standard short context cached input is priced at $0.5 per 1M tokens.
Read ai-costsPrompt caching reduces API expenses by charging a lower 'Read' rate for reused context after an initial 'Write' fee. For high-volume tasks, Haiku 4.5 offers the lowest entry point at $0.10 per 1M read tokens.
Read ai-costsMistral API pricing is based on token volume, with costs ranging from $0.11 per million input tokens for Mistral Small to $3.18 per million for Mistral Medium. Output tokens are priced higher across all tiers, peaking at $9.50 per million tokens on the Medium model.
Read ai-costsGemini API costs for current Flash models (3.6, 3.7, and 3.8) are $0.75 per 1 million input tokens and $3.75 per 1 million output tokens on the Standard tier. Choosing between Batch, Standard, and Priority tiers allows teams to balance per-token expenses against processing speed and throughput.
Read ai-costsHeyGen pricing starts at $24 per month for Creator plans and $149 per month for the Business tier, with additional seats costing $20 monthly. A Business seat pays for itself when a user saving $75/hour reduces their manual workload by just 18 minutes per month.
Read