ZMIME
DeepSeek · Released Jan 20, 2025

DeepSeek-R1

Classic open reasoning model for transparent math, coding, and deliberate problem solving.

  • Open weights
  • Reasoning
  • Tool calling
Capability
138.6ECI · #118 of 158
Input
$1.02per 1M tokens
Output
$5.14per 1M tokens
Context
128K33K max output
Pricing

What it costs

Input
$1.02per million tokens
Output
$5.14per million tokens
Blended (3:1)
$2.05pricier than 73% of priced models

Typical monthly bills

Monthly usageEstimated cost
Side project2M input + 0.5M output tokens$4.62
Team assistant25M input + 5M output tokens$51.33
Production app250M input + 50M output tokens$513.28

Median of 12 paid API providers on models.dev.

Capability

Independent benchmarks

138.6Epoch Capabilities Index
#118 of 158 scored models
68Median 147174

The shaded band is Epoch AI’s confidence range (136.7–140.4); the tick marks the median scored model.

  • GPQA DiamondGraduate-level science questions71.7%
  • OTIS Mock AIME 2024–2025Competition mathematics53.3%

Scores from Epoch AI, run independently of DeepSeek.

Specifications

The details

Lab
DeepSeek
Released
Jan 20, 2025
Knowledge cutoff
Jul 2024
Context window
128,000 tokens
Max output
32,768 tokens
Inputs
Text
Output
Text
Reasoning
Always on
Tool calling
Yes
Structured output
No
Availability
13 API providerslisted on models.dev
Lab-reported

What DeepSeek claims

Published by the lab at launch. Settings vary, so compare these only with care.

BenchmarkScoreSettingSource
Aider Polyglot56.9 percent correct—Source
Artificial Analysis Coding Index15.9 index—Source
SciCode35.7 percent correct—Source
Terminal-Bench Hard6.1 success rate—Source
Same lab

More from DeepSeek

Questions

About DeepSeek-R1

How much does DeepSeek-R1 cost?

DeepSeek-R1 costs $1.02 input / $5.14 output per million tokens (median across 12 API providers). At a 3:1 input-to-output mix that is $2.05 per million tokens, more expensive than 73% of the 469 priced models we track.

What is the context window of DeepSeek-R1?

DeepSeek-R1 accepts up to 128,000 tokens per request and can write up to 32,768 tokens in one response.

How good is DeepSeek-R1?

Epoch AI gives DeepSeek-R1 a Capabilities Index score of 138.6 (likely range 136.7–140.4), ranking it #118 of 158 models Epoch has scored. Epoch AI benchmark results: GPQA Diamond 71.7%, OTIS Mock AIME 2024–2025 53.3%.

Is DeepSeek-R1 open source?

Yes. DeepSeek publishes the weights, so it can be downloaded and self-hosted.

What inputs does DeepSeek-R1 support?

DeepSeek-R1 accepts text and replies in text. It is a reasoning model, supports tool calling.

When was DeepSeek-R1 released?

DeepSeek released DeepSeek-R1 on Jan 20, 2025. Its training data runs to Jul 2024.

What are the best alternatives to DeepSeek-R1?

The closest current models from other labs on capability, price and release date are GPT-4.1 (OpenAI, ECI 136.2, $2.00 / $8.00), Claude Sonnet 3.7 (Anthropic, ECI 141.1, $3.00 / $15.00), Gemini 2.5 Flash Preview 05-20 (Google, ECI 141.6, $0.142 / $3.33) and Kimi K2 Instruct (Moonshot AI, ECI 141.3, $0.57 / $2.30).