Ctrl K
Back to models

DeepSeek

DeepSeek-V4-Pro-0813

  • Released
  • Open weights
  • MIT
Compare
Intelligence Index

Max setting

36.0#28 of 83measured models
Measured · Artificial Analysis v4.3.2
Context window
1.05M tok

Performance profile

ReasoningRank #9 of 107
92.8%
GPQA Diamond
CodingRank #31 of 94
51.0%
SciCode
Cost efficiencyRank #36 of 67
53.4pts / $
Intelligence ÷ eval cost

Capabilities

Context window

Modalities

Reasoning modes

Avg reasoning tokens / task — Artificial Analysis Intelligence Index v4.3.2: weighted average reasoning tokens per benchmark task, not a token limit. Source (opens in a new tab)

Integration

Reasoning trade-offs

Source (opens in a new tab)
Intelligence Index v4.3.2
DeepSeek-V4-Pro-0813
Cost per task (USD)
DeepSeek-V4-Pro-0813Max
Intelligence Index
36.0
Cost / task
$0.674
Evaluation time
10.0m

Pricing & access

API route
DeepSeekAPI
Tier & context

USD per 1M tokens

Input
$1.32
Output
$3.96
Cached input
$0.044

Run locally

Original download
892.7 GB
App memory
Unverified
Documented software
vLLMSetup guide (opens in a new tab)
Find GGUF downloads (opens in a new tab)Original weights (opens in a new tab)

Benchmarks

Full leaderboardSource (opens in a new tab)

Humanity’s Last Exam

Frontier reasoning

DeepSeek-V4-Pro-0813 · Max#24 of 126

  1. 1Claude Opus 5.5Max61.4%
  2. 2Claude Fable 5.1Max59.1%
  3. 3Gemini 4 ArgonHigh57.1%
  5. 21GPT-5.3-CodexExtra high42.5%
  6. 22Qwen3.8 2.4T A95BReasoning42.4%
  7. 23GLM-5.3Max42.3%
  8. 24DeepSeek-V4-Pro-0813Max41.0%
  9. 25GLM-5.3-FlashMax39.9%
  10. 26Ling-3.1-flashReasoning39.4%
  11. 27DeepSeek-V4.1-FlashMax39.2%

Peers use their best measured result on this benchmark.

Alternatives

Each model at its highest-scoring measured setting · Intelligence Index v4.3.2 and its evaluation cost per task.