AiCostCompare

DeepSeek V4 Pro

by DeepSeek · deepseek/deepseek-v4-pro · #131 cheapest of 325 paid models

Prices updated Jul 22, 2026, 1:31 AM UTC · refreshed hourly

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,...

Compare vs:Claude Sonnet 5Claude Opus 4.6Claude Haiku 4.5GPT-5.6 Sol

Pricing

Input

$0.435

per 1M tokens

Output

$0.87

per 1M tokens

Blended (3:1)

$0.544

3 input : 1 output

Cache read

$0.0036

per 1M cached tokens

Cache write

per 1M tokens

Cache & batch economics

Effective prices for DeepSeek V4 Pro with prompt caching.

Cache hit rateEffective blended $/1MExample request*vs no cache
0%$0.544$0.0452
50%$0.382$0.0237−48%
90%$0.253$0.006416−86%

*Example: 100K sticky context + 2K new input + 1K output. Batch discount is an approximate 50% off list rates (available for OpenAI / Anthropic / Google — not flagged for this creator). Confirm on the vendor’s pricing page.

What a request costs

WorkloadInput tokensOutput tokensCost / requestCost / 1K requests
Short chat message500300$0.000478$0.4785
Document summary8,0001,000$0.00435$4.35
Codebase question (RAG)30,0002,000$0.0148$14.79
Long-context analysis150,0005,000$0.0696$69.60

Performance

Intelligence Index

44.3

AA composite quality

Coding Index

59.4

Math Index

Agentic Index

36.4

Output speed

74 tok/s

median

Time to first token

0.99s

median

Time to first answer

60.41s

after reasoning tokens

Artificial Analysis benchmarks

BenchmarkScore
GPQA Diamond88.8%
Humanity's Last Exam35.9%
SciCode50.0%
IFBench76.5%
AA-LCR (Long Context Reasoning)66.3%
Terminal-Bench Hard46.2%
Terminal-Bench 2.164.0%
τ²-Bench (Telecom)96.2%
τ³-Bench Banking25.8%

Design Arena Elo

CategoryElo
Website1,261
Web apps1,005
Full-stack948
UI components1,263
Data viz1,212
SVG1,183
3D1,310
Game dev1,288
Godot game dev1,059
Code categories1,272
ASCII art1,194

Specs

Context window

1.05M

tokens

Max output

384K

tokens

Input modalities

text

Tokenizer

DeepSeek