TokenTally

DeepSeek cost planning

DeepSeek Reasoner token costs

Reasoner unlocks long chain-of-thought outputs. Model your prompt + CoT budgets before rolling it out.

Last pricing check: Mar 13, 2026

$0.55 per 1M prompt tokens$2.19 per 1M completion tokens64,000 token context

Why teams choose this model

Math + logic assistants
Complex troubleshooting copilots
Multi-step code analyzers
Open main calculator

Scenario planning

Realistic cost examples

Numbers use DeepSeek Reasoner V3.2 pricing

Advanced math tutor

Students walk through proofs with detailed reasoning chains.

Per request

$0.0039

Per month

$19.72

Tokens sent

12,000,000

800 prompt tokens1600 completion tokens5,000 requests/mo

Code explanation bot

Engineers ask for deep dives on complex repos.

Per request

$0.0037

Per month

$13.04

Tokens sent

9,100,000

1200 prompt tokens1400 completion tokens3,500 requests/mo

Incident investigator

Ops teams replay timeline data and get structured CoT write-ups.

Per request

$0.0048

Per month

$10.49

Tokens sent

7,260,000

1500 prompt tokens1800 completion tokens2,200 requests/mo

Compare with

DeepSeek

DeepSeek Chat Pricing Guide

Updated Mar 13, 2026

Anthropic

Claude 3.7 Sonnet Pricing Guide

Updated Mar 13, 2026

Anthropic

Claude Sonnet 4.6 Cost Planner

Updated Mar 13, 2026

FAQs

Why is Reasoner pricier?

It allocates extra tokens for Chain-of-Thought (up to 32K), so outputs are longer. Cache-hit inputs still drop to $0.14/M tokens.

How do I cap CoT tokens?

Use the max_output_tokens field in your API call. TokenTally's scenarios show how that affects cost.

Can I mix Chat and Reasoner?

Yes—run Chat for simple steps and escalate to Reasoner when you need deeper thinking.

Pricing sources