Independent AI index
for professionals

Home / Models / DeepSeek

DeepSeek · Language model

DeepSeek

Open-weight (MIT) Chinese language model with a 1M context window and a built-in reasoning mode, at very low API costs.

1475#8 of 12 in the index
Elo, independently measuredQuality from blind duels on LMArena: users choose between two anonymous answers. A difference of around 20 points is noticeable; a few points is noise. How we measure
−2 Eloin 30 days
Editorial scoreOur own assessment on 10 of suitability for professional work, price and privacy. This is a judgement, not a measurement; that is why it always sits alongside the Elo. How we measure
7,4
GDPR profileOur summary of what the provider itself publishes about your data: EU hosting possible, no training on your data for business use, and self-hosting available. Caution applies to a model with a documented problem, such as an investigation by a regulator. How we measure
GDPR: Caution
Entry price
Freepaid from $0.15 / 1M in
Context
1M tokens
Speed
153 tokens/sec
Usage on OpenRouter
35.8% of tokens

Quality over time

121 daily measurements

The line follows the strongest variant of the family; the figure at the top is the measured flagship.

Bron: LMArena · opgehaald 9 november 2026

Your data

GDPR: Caution
–Hosting in the EU possible
–No training on your data (business)
✓Self-hostable (open weights)

The DeepSeek service stores data in China and uses it for training by default (with an opt-out). Only an EU self-hosted variant of the open weights is defensible for sensitive data.

GDPR check by profession

The DeepSeek models

DeepSeek is a family. Choose the heaviest for difficult tasks, the fastest for volume.

ModelTypeEloPrice inPrice outContextBuzz
DeepSeek-V4 ProMost powerful1464$0.96$1.911.048.576105
DeepSeek-V4 FlashFast & cheap1439$0.09$0.181.048.576321
DeepSeek-V4.1 Flash 1017Newest–$0.01$0.51.048.57663
DeepSeek-V4.1 FlashFast & cheap (previous)–$0.05$0.571.048.576298
DeepSeek-V4.1 Flash MaxMax reasoning––––5
DeepSeek-V4 Pro 0813High variant–$0.11$4.441.048.57634
DeepSeek-V4 Flash HighHigh variant––––1

Bron: LMArena, OpenRouter, Hacker News · opgehaald 9 november 2026

Subscriptions

Chat / appFreeOn chat.deepseek.com
Self-hostingFreeOpen weights, your own compute costs
API (V4.1-Flash)$0.15 / 1M in$0.60 / 1M out (off-peak rate); peak hours (Mon–Fri 01–04 and 06–10 UTC) at double the price; cache-hit input $0.003 per 1M
API (V4-Pro)$0.66 / 1M in$1.98 / 1M out (off-peak rate); peak hours at double the price; cache-hit input $0.022 per 1M

Where it excels

  1. 1Very low API costs
  2. 2Open weights (MIT), self-hostable
  3. 3Strong at reasoning, maths and code
  4. 4Large context window (1M tokens)

DeepSeek by profession

4 professions where we recommend it, listed first.

The full index on one scale

Ordered by quality, height is our editorial score. DeepSeek is highlighted.

8,69,07,68,98,37,97,77,47,88,48,07,6
#1Gemini#2Claude#3Kimi#4ChatGPT#5Copilot#6Qwen#7GLM (Z.ai)#8DeepSeek#9Grok#10Le Chat (Mistral)#11Perplexity#12Llama

Scale 0 to 10 · editorial score

1Gemini8,62Claude9,03Kimi7,64ChatGPT8,95Copilot8,36Qwen7,97GLM (Z.ai)7,78DeepSeek7,49Grok7,810Le Chat (Mistral)8,411Perplexity8,012Llama7,6

Compare DeepSeek with another

DeepSeek vs GeminiDeepSeek vs ClaudeDeepSeek vs KimiDeepSeek vs ChatGPTDeepSeek vs CopilotDeepSeek vs QwenDeepSeek vs GLM (Z.ai)DeepSeek vs GrokDeepSeek vs Le Chat (Mistral)DeepSeek vs PerplexityDeepSeek vs Llama

What the community says

DeepSeek V4 Pro 08131041 points · 453 commentsDeepSeek v4.1 Flash1016 points · 574 commentsDeepSeek-v4.1-Flash-1017959 points · 727 commentsDeepSeek: DeepSeek v4.1 Flash 1017520 points · 5 comments

Hacker News, last 90 days, praise and criticism.

Tools that run on DeepSeek

Inworld AIGame development