TokenTally

DeepSeek cost planning

DeepSeek Chat token costs

See how DeepSeek Chat's cache-aware pricing impacts high-volume assistants, ops copilots, and QA bots.

Last pricing check: Mar 13, 2026

$0.27 per 1M prompt tokens$1.10 per 1M completion tokens64,000 token context

Why teams choose this model

Workflow copilots that fire thousands of short prompts
Batch content clean-up with shared system prompts
Analytics assistants embedded in dashboards
Open main calculator

Scenario planning

Realistic cost examples

Numbers use DeepSeek Chat V3.2 pricing

Cache-friendly support queue

Tier-1 bot reusing the same system instructions all day.

Per request

$0.0003

Per month

$14.76

Tokens sent

27,000,000

400 prompt tokens200 completion tokens45,000 requests/mo

Batch newsletter digest

Weekly digest assembled from product notes and changelogs.

Per request

$0.0008

Per month

$1.43

Tokens sent

2,520,000

900 prompt tokens500 completion tokens1,800 requests/mo

Ops analyst assistant

Employees ask natural-language questions over metrics.

Per request

$0.0005

Per month

$6.17

Tokens sent

11,040,000

600 prompt tokens320 completion tokens12,000 requests/mo

Compare with

DeepSeek

DeepSeek Reasoner Cost Calculator

Updated Mar 13, 2026

Anthropic

Claude 3.7 Haiku Cost Calculator

Updated Mar 13, 2026

Anthropic

Claude Haiku 4.5 Pricing Guide

Updated Mar 13, 2026

FAQs

How does the cache-hit discount work?

DeepSeek automatically caches shared prompt prefixes. Repeat system instructions cost $0.07/M tokens instead of $0.27/M. No extra headers needed.

What should I budget for outputs?

Outputs are always billed at $1.10 per million tokens, so longer responses dominate the bill if prompts stay cached.

Is the API OpenAI-compatible?

Yes—switch the base URL to api.deepseek.com and drop in your DeepSeek key. TokenTally lets you compare before migrating.

Pricing sources