# DeepSeek-V2

> The 236B Mixture-of-Experts model that started China's LLM price war

- Maker: DeepSeek
- Line: DeepSeek V3
- Canonical page: https://ai-tldr.dev/models/deepseek-v2/
- Updated: 2026-06-18

## In short

DeepSeek-V2 is an open-weight Mixture-of-Experts large language model released by Chinese AI lab DeepSeek in May 2024. It has 236 billion total parameters but activates only 21 billion per token, which is what lets a model this large run cheaply. It has a 128K-token context window. You can use it through DeepSeek and Hugging Face.

## Facts

| Fact | Value | Source |
| --- | --- | --- |
| Context window | 128K tokens | [1] |
| Licence | DeepSeek Model License (source-available; commercial use permitted). Repository code is MIT-licensed. · open weights | [1] |
| Released | May 2024 | [2] |
| Parameters | 236B total, 21B activated per token (Mixture-of-Experts) | [1] |
| Knowledge cutoff | Not officially published by DeepSeek | [1] |

## Sources

1. [DeepSeek-V2 on Hugging Face](https://huggingface.co/deepseek-ai/DeepSeek-V2)
2. [DeepSeek (Wikipedia) — history & pricing](https://en.wikipedia.org/wiki/DeepSeek)

## Overview

DeepSeek-V2 is an open-weight Mixture-of-Experts large language model released by Chinese AI lab DeepSeek in May 2024. It has 236 billion total parameters but activates only 21 billion per token, which is what lets a model this large run cheaply. It supports a 128K-token context window and was pretrained on 8.1 trillion tokens of text and code.

Its two headline ideas are DeepSeekMoE and Multi-head Latent Attention (MLA). DeepSeekMoE splits the feed-forward layers into 2 shared experts plus 160 routed experts and uses just 6 of them per token, so most of the network sits idle on any given step. MLA compresses the key-value cache that normally dominates inference memory, and DeepSeek reports it cuts that cache by 93.3% and boosts maximum throughput 5.76x compared with the older dense DeepSeek 67B. Together these techniques are why DeepSeek-V2 could be served so cheaply.

DeepSeek-V2 is best remembered for triggering a price war: its launch API rates were so low (the Financial Times reported roughly 2 RMB per million output tokens) that other Chinese labs quickly cut their own prices. The model is now discontinued — DeepSeek replaced it with DeepSeek-V2.5 in September 2024 and the much larger DeepSeek-V3 in December 2024 — but its open weights remain on Hugging Face and its MLA + MoE recipe carried directly into those successors.

## Benchmarks

- MMLU (Base): 78.5% ([source](https://huggingface.co/deepseek-ai/DeepSeek-V2))
- BBH (Base): 78.9% ([source](https://huggingface.co/deepseek-ai/DeepSeek-V2))
- C-Eval (Base): 81.7% ([source](https://huggingface.co/deepseek-ai/DeepSeek-V2))
- CMMLU (Base): 84% ([source](https://huggingface.co/deepseek-ai/DeepSeek-V2))
- GSM8K (Chat): 92.2% ([source](https://github.com/deepseek-ai/DeepSeek-V2))
- HumanEval (Chat): 81.1% ([source](https://github.com/deepseek-ai/DeepSeek-V2))
- MATH (Chat): 53.9% ([source](https://github.com/deepseek-ai/DeepSeek-V2))
- MMLU (Chat): 77.8% ([source](https://github.com/deepseek-ai/DeepSeek-V2))

## Strengths

- Extremely cheap to serve for its size — only 21B of 236B parameters activate per token
- Multi-head Latent Attention cuts KV-cache memory by 93.3%, enabling long contexts at low cost
- Strong coding and math scores for an open model of its era (HumanEval 81.1, GSM8K 92.2 on the Chat model)
- 128K-token context window with reliable long-context retrieval
- Open weights under a commercially-permissive license, with an MIT-licensed code repository
- Strong bilingual (Chinese + English) performance — C-Eval 81.7, CMMLU 84.0 on the base model

## Best for

- Low-cost, high-throughput text generation and chat at scale
- Coding assistance and code generation
- Math and reasoning tasks
- Long-document question answering and summarization within a 128K window
- Chinese and English bilingual applications
- Self-hosting on your own GPUs (e.g. via vLLM) when you need open weights and data control
- A historical reference / baseline for studying MoE and MLA architectures

## How to access

- [DeepSeek](https://api-docs.deepseek.com/quick_start/pricing) — model id `deepseek-chat (historical; endpoint has since moved to newer models)`
- [Hugging Face](https://huggingface.co/deepseek-ai/DeepSeek-V2) — model id `deepseek-ai/DeepSeek-V2`

## DeepSeek V3 versions

- [DeepSeek-V3.2](https://ai-tldr.dev/models/deepseek-v3-2/) (1 December 2025)
- [DeepSeek-V3.2-Speciale](https://ai-tldr.dev/models/deepseek-v3-2-speciale/) (1 December 2025)
- [DeepSeek-V3.2-Exp](https://ai-tldr.dev/models/deepseek-v3-2-exp/) (29 September 2025)
- [DeepSeek-V3.1-Terminus](https://ai-tldr.dev/models/deepseek-v3-1-terminus/) (22 September 2025)
- [DeepSeek-V3.1](https://ai-tldr.dev/models/deepseek-v3-1/) (21 August 2025)
- [DeepSeek-V3-0324](https://ai-tldr.dev/models/deepseek-v3-0324/) (24 March 2025)
- [DeepSeek-V3](https://ai-tldr.dev/models/deepseek-v3/) (26 December 2024)
- [DeepSeek-V2.5](https://ai-tldr.dev/models/deepseek-v2-5/) (5 September 2024)
- DeepSeek-V2 (May 2024) — this page

## FAQ

### What is DeepSeek-V2?

DeepSeek-V2 is an open-weight Mixture-of-Experts large language model released by DeepSeek in May 2024. It has 236 billion total parameters but activates only 21 billion per token, supports a 128K-token context window, and was trained on 8.1 trillion tokens.

### Is DeepSeek-V2 still available?

Its open weights are still downloadable on Hugging Face, but the model is discontinued. DeepSeek replaced it with DeepSeek-V2.5 in September 2024 and DeepSeek-V3 in December 2024, and the hosted API now serves newer models.

### What made DeepSeek-V2 important?

Two things. Technically, it introduced Multi-head Latent Attention (MLA) and the DeepSeekMoE design, cutting KV-cache memory by 93.3% and boosting throughput 5.76x versus the dense DeepSeek 67B. Commercially, its very low launch price set off a price war among Chinese AI labs.

### How big is DeepSeek-V2 and how much runs at once?

It has 236B total parameters but only 21B activate per token because it is a Mixture-of-Experts model. Each MoE layer uses 2 shared experts plus 160 routed experts and selects just 6 of them per token, which keeps inference cheap despite the large total size.

## Links

- [DeepSeek-V2 paper (arXiv 2405.04434)](https://arxiv.org/abs/2405.04434)
- [DeepSeek-V2 on Hugging Face](https://huggingface.co/deepseek-ai/DeepSeek-V2)
- [DeepSeek-V2 GitHub repository](https://github.com/deepseek-ai/DeepSeek-V2)
- [DeepSeek API pricing docs](https://api-docs.deepseek.com/quick_start/pricing)
- [DeepSeek (Wikipedia) — history & pricing](https://en.wikipedia.org/wiki/DeepSeek)

Source: AI/TLDR LLM registry, https://ai-tldr.dev/models/deepseek-v2/