DeepSeek Reasoning & Thinking Live Rates • September 2026

DeepSeek R1 Pricing & Token Economics

May 28th update to the original DeepSeek R1 Performance on par with OpenAI o1, but open-sourced and with fully open reasoning tokens.

Context Window 164K
Max Completion 33K Tokens
Inference Speed Deliberate / Thinking
CostRatio Score A+

Normalized Token Pricing Matrix

Real-time input, completion, and cache rates normalized to $/1M tokens.

📥 Prompt Tokens
$ 0.50 / 1M tokens

Cost for prompts, instructions, document retrieval, and system context sent into the model.

~$0.0005 per 1K tokens
📤 Completion Tokens
$ 2.15 / 1M tokens

Cost for generated text, code, tool calls, and structured JSON responses returned by the model.

~$0.0022 per 1K tokens
⚡ Prompt Caching
$ 0.350 / 1M tokens

Massive 30% savings on repeated context, documents, and system instructions.

Cached Read Discount
⚖️ 3:1 Blended Benchmark
$ 0.912 / 1M tokens

Standard industry benchmark assuming 75% prompt reads and 25% completion generation.

Weighted Benchmark

DeepSeek R1 Monthly Bill Estimator

Estimated monthly costs across four standardized enterprise and developer workloads.

Workload Scenario Monthly Token Volume Standard Bill With Prompt Caching
💬
Customer Support Chatbot
10M prompt tokens + 2.5M output tokens per month
12.5M tokens (10M in / 2.5M out) $10.38/mo $9.32/mo -10%
📑
Document Analysis & RAG
30M prompt tokens (PDFs/context) + 3M output summaries
33.0M tokens (30M in / 3M out) $21.45/mo $18.30/mo -15%
💻
Coding Agent / IDE Swarm
60M prompt tokens (repo context) + 15M code completions
75.0M tokens (60M in / 15M out) $62.25/mo $55.95/mo -10%
🤖
Enterprise Agentic Workflow
120M prompt tokens (reasoning loops) + 30M output actions
150.0M tokens (120M in / 30M out) $124.50/mo $111.90/mo -10%

Need custom token inputs or prompt volumes?

Use our interactive multi-model token calculator to test exact prompt and completion sliders.

Open Interactive Token Calculator →

Architecture & Capabilities

Technical parameters, supported modalities, and optimal developer use cases.

🎯 Recommended Use Cases

Deep mathematical derivation, competitive coding, chain-of-thought logic

Supported Modalities

📝 Text 🛠️ Function / Tool Calling

📐 Context & Output Limits

  • Maximum Context Window: 163,840 tokens (164K)
  • Maximum Output Length: 32,768 tokens
  • Speed / Latency Tier: Deliberate / Thinking
  • Official Provider: DeepSeek

Frequently Asked Questions

Everything you need to know about DeepSeek R1 API pricing and token calculations.

How much does DeepSeek R1 cost per 1M tokens?

DeepSeek R1 charges $0.50 per 1M input tokens and $2.15 per 1M output tokens. For standard balanced workloads (3:1 input:output ratio), the effective blended rate is $0.912/1M tokens.

Does DeepSeek R1 support prompt caching?

Yes. Cached prompt reads are billed at $0.3500 per 1M tokens, saving you up to 30% on repeated context, RAG documents, and system instructions.

What is DeepSeek R1's maximum context window?

DeepSeek R1 supports up to 163,840 tokens (164K) in its context window, allowing you to ingest large files and multi-turn chat history with up to 32,768 tokens generated per response.

How does DeepSeek R1 compare to other models?

With a CostRatio rating of A+, DeepSeek R1 provides competitive token economics for reasoning & thinking tasks. Review the peer comparison table above to compare input and completion rates directly against alternative models.

Explore all latest-generation models and interactive calculators

View Full AI Model Matrix →