LLM·Dex
Newsletter
Open weightsDeepSeekopentext

DeepSeek-R1

First open-weight reasoning model to match o1, the release that proved RL-from-scratch reasoning training was reproducible.

Updated

Quick facts

Released
Jan 2025
Context
128K tokens
Output / 1M
$2.19 / 1M tokens
License
MIT

About DeepSeek-R1

DeepSeek-R1's January 2025 release was a watershed moment, the first openly-trained reasoning model that matched OpenAI o1 on standard benchmarks. The release also included the R1-Zero ablation showing that RL-only post-training (no SFT) could produce reasoning behavior, which kicked off a wave of reasoning-from-scratch reproductions.

In 2026, R1 remains a top pick for reasoning-heavy open-weight workloads. Many production teams use it for math, science, and agent planning.

Benchmarks

Published scores from DeepSeek's model card or independent leaderboards. We do not publish numbers we cannot source, see methodology.

HumanEval
,
Python coding pass@1
MMLU
,
Broad academic knowledge
GPQA
71.5
Graduate-level reasoning
SWE-bench
,
Real software-engineering tasks
  • GPQA71.5

Capabilities

Strengths

  • Open-weight reasoning model on par with o1
  • MIT license
  • Cheap reasoning per token

Tracked weaknesses

  • Slow, reasoning is slow by design
  • No vision

Pricing

Per-million-token rates as published by DeepSeek.

TierPriceNotes
Input$0.55 / 1M tokensTokens you send to the model
Output$2.19 / 1M tokensTokens the model generates
Context128K tokensMax combined input + output

Call DeepSeek-R1 from your code

Drop-in snippet for the DeepSeek SDK. Set your API key in the environment and run.

typescript
import OpenAI from "openai";

const client = new OpenAI({
  // Use OPENAI_API_KEY for OpenAI, or your provider's key + baseURL.
  apiKey: process.env.OPENAI_API_KEY,
});

const completion = await client.chat.completions.create({
  model: "deepseek-r1",
  messages: [
    { role: "user", content: "What's the time complexity of quicksort?" },
  ],
});

console.log(completion.choices[0].message.content);

Best for

Tasks where DeepSeek-R1 ranks among LLMDex's top picks.

Compare DeepSeek-R1 with…

Frequently asked

  • How much does DeepSeek-R1 cost per million tokens?
    DeepSeek-R1 is priced at $0.55 / 1M tokens for input tokens and $2.19 / 1M tokens for output tokens via the official DeepSeek API at the time of writing.
  • What is DeepSeek-R1's context window?
    DeepSeek-R1 supports a context window of 128K tokens.
  • Is DeepSeek-R1 open source?
    DeepSeek-R1 ships with open weights under the MIT license. You can self-host it, fine-tune it, and (subject to the license terms) deploy it commercially.
  • When was DeepSeek-R1 released?
    DeepSeek-R1 was released on Jan 20, 2025 by DeepSeek.
  • What is DeepSeek-R1's knowledge cutoff?
    DeepSeek-R1's training data has a knowledge cutoff of Jul 2024. For information after that date you'll need a tool-use or web-search wrapper.
Friday digest

Intelligence, distilled weekly.

One short email every Friday, new model launches, leaderboard moves, and pricing drops. Curated by hand. Free, no spam.