Read dated DeepSeek V3.2 channel quotes, separate cache rates and calculate your workload. Distinguish this model from current DeepSeek aliases and direct-service pricing.
Explanation reviewed:
Match the exact model and channel
This page concerns DeepSeek V3.2, not every current DeepSeek model. The recorded OpenRouter channel is deepseek/deepseek-v3.2. Direct DeepSeek aliases and gateway identifiers can refer to different deployed versions; check the actual service documentation before using a price.
The direct DeepSeek pricing page reviewed on October 10 lists newer models, so its current table is not evidence of a V3.2 direct price. OpenRouter maintains a V3.2 page with provider-specific rates and routing conditions.
Read the quote date and validity
The table above is a dated stored snapshot, not a live quote fetch. Its source date, expiry and currency belong to that particular record. The editorial date below refers to this explanation and does not refresh the quote.
OpenRouter provider selection, temporary discounts, cache categories and payment charges can change total cost. A headline minimum is not a promise that every request will use that provider or price. Verify the source and checkout before spending.
Calculate the workload rather than the top-up
Apply the displayed input and output rates to the matching token counts. For 1M input and 0.2M output, the base estimate is one input rate plus 0.2 times the output rate. The interactive counts above let you compare this same workload.
The table’s base calculation excludes cache, tools and extra fees. If cache reads are included in your input total, split them out and use the cache rate with the full billing formula. Count billed retries and accepted results separately.
What this page can and cannot establish
Source-backed quotes help estimate cost. They do not establish latency, availability, tool compatibility or refund handling. Consult the provider evidence and a small authorized task test before relying on a route for production work.