modelgrep
D

DeepSeek: DeepSeek V4 Pro

deepseek/deepseek-v4-pro

43rd smartest of 180ReasoningToolsJSON
Use via OpenRouter ↗
Intelligence
39.3
43rd of 180
Design Elo
1348
3D
Speed
62
137th fastest
Latency
604ms
first token
Input price
$0.435
156th cheapest
Context
1.0M
384K max out

How it compares

Smarter than76%
of all ranked models
Faster than55%
of all ranked models
Cheaper than48%
of all ranked models

Overview

DeepSeek V4 Pro is a large-scale Mixture-of-Experts model from DeepSeek with 1.6T total parameters and 49B activated parameters, supporting a 1M-token context window. It is designed for advanced reasoning, coding,...

Benchmarks

independent · via OpenRouter
Artificial Analysis73th percentile
Intelligence Index
39.3
Coding Index
38.4
Agentic Index
63.3
GPQA Diamond
72%
Humanity's Last Exam
8%
SciCode
42%
Tau²-Bench (agentic)
91%
Design Arena · Elo7,314 tournaments
3D
1348
Game Dev
1322
codecategories
1297
Website
1285
UI Component
1282
Data Viz
1228
asciiart
1200
svg
1193
godotgamedev
1098
webapps
1031
fullstack
948

Providers & pricing (16)

ProviderIn $/MOut $/MUptime
DeepSeekcache$0.435$0.870100%
Baidufp8$0.760$1.5299.9%
StreamLake$0.870$1.7495.5%
GMICloudfp8$1.27$2.5493.2%
DeepInfrafp4$1.30$2.6094.4%
DigitalOcean$1.48$2.9695%
SiliconFlowfp8$1.60$3.1399.8%
Novitafp8$1.60$3.2098.9%
Alibaba$1.61$3.2299.7%
AtlasCloudfp8$1.68$3.3899.4%
Venice$1.73$3.8096.8%
Parasailfp8$1.74$3.48
Fireworks$1.74$3.4897.6%
Together$1.74$3.4895.4%
WandBfp8$1.74$3.4898.2%
NextBitfp8$1.80$3.20

Specifications

Context window1.0M
Max output384K
Knowledge cutoff
Input modalitiestext
Output modalitiestext
Prompt cachingSupported
Cache read price$0.0036/M
ModeratedNo

DeepSeek V4 Pro FAQ

How much does DeepSeek V4 Pro cost?

DeepSeek V4 Pro costs $0.435 per million input tokens and $0.870 per million output tokens via OpenRouter, making it 156th cheapest of 298 paid models.

How smart is DeepSeek V4 Pro?

DeepSeek V4 Pro scores 39.3 on the Artificial Analysis Intelligence Index, ranking 43rd of 180 benchmarked models, with a GPQA Diamond score of 72%.

How fast is DeepSeek V4 Pro?

DeepSeek V4 Pro generates around 62 tokens per second with 604ms time-to-first-token (p50), the 137th fastest tracked model.

What is DeepSeek V4 Pro's context window?

DeepSeek V4 Pro supports a 1.0M-token context window and can output up to 384K tokens. It accepts text input.

Compare head-to-head