DeepSeek calculator

Separate cache hits before you trust the DeepSeek bill.

Estimate DeepSeek V4 Flash and V4 Pro costs from cache-hit input, cache-miss input, output tokens, thinking mode, and daily request volume. The split matters because DeepSeek's cached input can be dramatically cheaper than fresh input.

Inputs

ModelThinking modeCalls per day

Cache-hit input tokensCache-miss input tokensOutput tokens

Days per month

DeepSeek reports cache hits as prompt_cache_hit_tokens and cache misses as prompt_cache_miss_tokens. Use API usage mode when you have those fields.

Result

$19.82estimated monthly cost

Per request<$0.01

Daily$0.66

Total input18,000

Monthly calls30,000

Cache hit input<$0.01

Cache miss input<$0.01

Output<$0.01

Annual$238

High confidence: based on DeepSeek usage fields such as prompt_cache_hit_tokens and prompt_cache_miss_tokens.

Pricing checked 2026-05-09. DeepSeek V4 Flash: cache hit $0.0028/1M, cache miss $0.14/1M, output $0.28/1M tokens.

Thinking output should be included in output token planning because DeepSeek bills by generated tokens.

Official model. The compatibility aliases deepseek-chat and deepseek-reasoner map to V4 Flash non-thinking and thinking modes.

References

Built around DeepSeek's cache-aware billing model.

DeepSeek Models & Pricing DeepSeek Token & Token Usage DeepSeek Context Caching DeepSeek Thinking Mode

DeepSeek context caching is enabled by default and reports cache status through usage fields. This static page does not call DeepSeek APIs or ask for API keys; it is a transparent planning calculator for pre-production budgeting.