DeepSeek

Also known as: DeepSeek-R1 · DeepSeek-V3

On this page 7

DeepSeek is a Chinese artificial intelligence company based in Hangzhou and the name of its family of large language models, best known for the general-purpose DeepSeek-V3 and the reasoning model DeepSeek-R1. Among AI models & assistants, DeepSeek stands out for releasing open-weight models that approach the performance of leading proprietary systems at a substantially lower reported training and inference cost.

The company was founded in 2023 with funding from the quantitative hedge fund High-Flyer, led by Liang Wenfeng. It also operates a free chat assistant on the web and in mobile apps, plus a paid developer API. The release of DeepSeek-R1 in January 2025 prompted wide debate about the cost of frontier AI and about export controls on advanced chips.

The DeepSeek model family and its releases

DeepSeek has released models at a rapid, incremental pace, with each generation introducing an architectural or training change that later releases reused. The table below lists the main milestones; parameter counts are those published by the company.

ModelReleasedSizeMain contribution
DeepSeek LLMLate 20237B and 67B, denseFirst general open models
DeepSeek-CoderLate 2023Up to 33B, denseCode-specialized training
DeepSeek-V2May 2024236B total, 21B activeMulti-head latent attention, sparse experts
DeepSeek-V3December 2024671B total, 37B activeFP8 training, multi-token prediction
DeepSeek-R1January 2025671B total, 37B activeReasoning learned through reinforcement
DeepSeek-V3.1 and V3.22025671B classHybrid thinking modes, sparse attention
Major DeepSeek model releases

Alongside the flagship models, the company published distilled R1 variants from 1.5B to 70B parameters, built by fine-tuning Qwen and Llama base models on reasoning traces generated by DeepSeek-R1. These smaller models run on a single workstation GPU and made local reasoning models widely accessible.

Architecture and efficiency techniques

DeepSeek’s efficiency comes mainly from a mixture-of-experts (MoE) architecture, in which only a small subset of parameters is used for each token. In DeepSeek-V3, roughly 37 billion of 671 billion parameters are active per token, so inference compute resembles that of a much smaller dense model while total knowledge capacity stays large.

Several further techniques reduce memory and compute. The following table summarizes the most cited ones and what each achieves.

TechniqueWhat it doesPractical effect
DeepSeekMoEMany fine-grained experts plus shared expertsBetter specialization per active parameter
Multi-head latent attentionCompresses keys and values into a latent vectorMuch smaller key-value cache
Auxiliary-loss-free load balancingAdjusts routing bias instead of adding a penaltyEven expert use without hurting quality
Multi-token predictionTrains the model to predict several future tokensDenser training signal, faster decoding
FP8 mixed precisionRuns most training math in 8-bit floating pointLower memory and faster training
DeepSeek Sparse AttentionAttends only to selected earlier tokensCheaper long-context processing
Key DeepSeek efficiency techniques

DeepSeek reported that the final training run of DeepSeek-V3 used about 2.8 million H800 GPU hours, which it priced at roughly 5.6 million US dollars. This figure excludes research, failed experiments, data work and hardware ownership, so it is not the total cost of building the model, a distinction often lost in public discussion.

How DeepSeek-R1 learned to reason

DeepSeek-R1 is a reasoning model that produces a long chain of thought before giving its final answer, in the same class as the OpenAI o-series. Its notable feature is the training recipe, which relied heavily on large-scale reinforcement learning with automatically checkable rewards.

The precursor, DeepSeek-R1-Zero, was trained with reinforcement learning directly on the V3 base model, without supervised examples of reasoning. Rewards came from rule-based checks, such as whether a math answer matched or code passed tests. The algorithm, Group Relative Policy Optimization (GRPO), compares several sampled answers to the same prompt and avoids a separate critic model, which saves memory.

R1-Zero showed emergent behaviors such as self-verification, but its outputs were often hard to read and mixed languages. The final R1 therefore added a small cold-start supervised stage, further reinforcement learning and a broader fine-tuning round.

StageData or signalPurpose
Cold-start fine-tuningThousands of curated reasoning examplesReadable chain-of-thought format
Reasoning reinforcementRule-based correctness rewardsStronger math, code and logic
Rejection sampling and tuningFiltered model outputs plus general dataBroader skills beyond reasoning
Final reinforcementCorrectness and preference rewardsHelpfulness and harmlessness
DeepSeek-R1 training stages

DeepSeek compared with other model families

DeepSeek is usually compared with other open-weight families such as Llama and Qwen, and with proprietary assistants such as Claude, Gemini and GPT-5. Its distinctive position combines frontier-level reasoning scores, a permissive license and low API prices.

AspectDeepSeekTypical proprietary modelTypical open-weight model
WeightsPublished, MIT license for R1Not publishedPublished, license varies
ArchitectureLarge sparse MoEMostly undisclosedDense or MoE
Self-hostingPossible but hardware-heavyNot possibleOften practical
API pricingLow per million tokensMedium to highDepends on host
ModalitiesMainly textText, image, audioMostly text
DeepSeek against typical alternatives

Running the full 671B model locally requires several hundred gigabytes of accelerator memory, even with quantization, so most organizations either use a cloud host or choose a distilled variant.

Limitations, censorship and data concerns

DeepSeek models carry content restrictions shaped by Chinese regulation. The hosted assistant declines or deflects questions on politically sensitive topics, and similar behavior appears to varying degrees in the open weights. Organizations should test for these biases before deploying the models in public-facing products.

  • Data residency: the hosted app and API process and store data on servers in China, which has led several governments and agencies to restrict use on official devices.
  • Reasoning overhead: R1 generates long thinking traces, increasing latency and token cost for simple tasks.
  • Safety testing gaps: independent red-team evaluations have reported weaker refusal of harmful requests than leading proprietary models.
  • Limited multimodality: the main chat models focus on text, while vision support has come through separate releases.

Self-hosting the open weights removes the data residency concern, because no prompts leave the organization, but it does not remove the trained-in content behavior.

The idea in brief

DeepSeek is a Chinese company that makes AI chatbots and the models behind them. Its models can write, answer questions, solve math problems and write code. Many people use its free chat app.

What made DeepSeek famous is that it reached near top-level quality at a lower cost. It also gave away the model files so anyone can download and run them. Think of a carmaker that builds a car almost as fast as the best, sells it cheaply and publishes the engine design. The catch is that the hosted service keeps data in China and avoids some political topics.

A practical scenario

A mid-sized engineering firm wants an internal assistant that helps staff check calculations and summarize technical reports. Its legal team will not allow confidential documents to be sent to an external service. The IT department therefore deploys a distilled DeepSeek-R1 model with 32 billion parameters on two in-house GPU servers, quantized to 4-bit precision.

In a pilot with 40 engineers, the model handles unit conversions and step-by-step load calculations well, though answers take 20 to 60 seconds because of long reasoning traces. The team adds a fast non-reasoning model for simple questions and routes only calculation tasks to R1. The result is useful reasoning support with no data leaving the building, at a hardware cost the firm could justify after three months of use.