Private AI

Nano GPT logo
NanoGPT
Nano GPT logo
NanoGPT
Nano GPT logo
NanoGPT
Search
 
 
 
 
All conversations

Create

Chat
Studio

Explore

Models
Pricing
AI Apps
Blog

Library

Gallery
Workspace
Usage & billing
API
Suggestions and bugs
Use cases

|

Terms

|

Privacy

|

Refunds
Usage & billing
Teams
Referrals
Presets
Settings
FAQPress & mediaBrand assetsBenchmarksNano RPCEarn crypto
  1. Models
  2. Text
  3. DeepSeek

DeepSeek V4 Pro (April preview)

DeepSeek V4 Pro April preview (2026-04-23), the first V4 Pro checkpoint: a large-scale Mixture-of-Experts model with a 1M-token context window for reasoning, coding, long-horizon agent workflows, knowledge, math and software engineering. Plain DeepSeek V4 Pro now means V4 Pro 0813.

  • Reasoning
  • Tool Calling
  • Structured Output

Added Apr 24, 2026

Model weights

Pricing

Auto routing · per 1M tokens
Input
$0.87
Output
$1.78
Cache read
$0.0796
Compare provider prices

Specifications

Context window
1M
Max output
384K
Parameters
1.6T / 49B
Total / active

Benchmarks

Sourced from LMArena.

Arena Score

1458.1

Overall Rank

#65 / 421

Votes

56,906

Confidence Interval

1454.2 - 1462.0

Category Scores

Coding

#80 / 416

16,930 votes

1500.6

Math

#88 / 404

3,019 votes

1443.6

Longer Query

#60 / 399

25,531 votes

1472.2

Creative Writing

#48 / 419

9,595 votes

1446.7

Instruction Following

#61 / 421

19,767 votes

1453.4

Hard Prompts

#67 / 421

38,011 votes

1480.5

Additional Categories
22

Japanese

#41 / 288

716 votes

1454.5

French

#49 / 301

2,190 votes

1484.8

German

#49 / 320

985 votes

1467.7

Multi Turn

#51 / 419

10,267 votes

1475.1

Industry Legal And Government

#52 / 393

4,431 votes

1474.5

Polish

#55 / 244

1,119 votes

1464.0

Industry Writing And Literature And Language

#56 / 420

13,947 votes

1451.3

Industry Entertainment And Sports And Media

#60 / 419

12,613 votes

1436.0

Industry Business And Management And Financial Operations

#61 / 414

11,255 votes

1459.8

Russian

#62 / 385

6,253 votes

1460.9

Industry Life And Physical And Social Science

#63 / 419

9,018 votes

1479.7

Industry Medicine And Healthcare

#63 / 390

3,999 votes

1477.5

Spanish

#65 / 303

1,711 votes

1456.5

Exclude Ties

#66 / 421

43,105 votes

1462.1

English

#67 / 421

25,016 votes

1467.6

Korean

#67 / 289

1,057 votes

1415.0

Non English

#67 / 421

31,890 votes

1444.8

Hard Prompts English

#68 / 419

17,634 votes

1485.3

Industry Software And It Services

#72 / 421

23,353 votes

1491.5

Industry Mathematical

#73 / 401

3,208 votes

1459.9

Expert

#79 / 371

5,750 votes

1479.5

Chinese

#81 / 395

3,321 votes

1487.8

Published 2026-11-09 · Matched as DeepSeek-V4-Pro · CC BY 4.0

LMArena Dataset

Providers

Auto routing is available for this model. Explicit provider selection is not available.

Auto
Auto
Recommended
$0.87$1.78$0.0796N/AN/A
Batch API
ProviderInputOutputCached input
Auto
Autodeepseek/deepseek-v4-pro-20260423
$0.87/M$1.78/M$0.0796/M

Batch rows for this model use normal routing and cost the same as standard API requests. Set provider in a row to choose a provider.

Reliability

Related text models

Compare DeepSeek V4 Pro (April preview) with similar models from the same provider or model family.

DeepSeek V4 Pro (April preview, Thinking)

deepseek/deepseek-v4-pro-20260423:thinking

DeepSeek V4 Pro April preview (2026-04-23) with chain-of-thought enabled: a large-scale Mixture-of-Experts model with a 1M-token context window. Plain DeepSeek V4 Pro now means V4 Pro 0813.

DeepSeek V4 Pro 0813

deepseek/deepseek-v4-pro-0813

DeepSeek V4 Pro 0813 is the general-availability release of DeepSeek V4 Pro, built for coding, tool use, cybersecurity, automation, and long-horizon agent workflows. DeepSeek reports strong gains over the preview and scores above Opus 4.8 on Terminal Bench 2.1, Cybergym, DeepSWE, and AutomationBench. It supports a 1M-token context window.

DeepSeek V4 Pro 0813 Thinking

deepseek/deepseek-v4-pro-0813:thinking

DeepSeek V4 Pro 0813 Thinking enables reasoning by default on the general-availability release of DeepSeek V4 Pro, built for coding, tool use, cybersecurity, automation, and long-horizon agent workflows. DeepSeek reports strong gains over the preview and scores above Opus 4.8 on Terminal Bench 2.1, Cybergym, DeepSWE, and AutomationBench. It supports a 1M-token context window.

DeepSeek V4.1 Flash 0910 Uncensored

deepseek/deepseek-v4.1-flash-0910-uncensored

Uncensored DeepSeek V4.1 Flash with text and image input, optional reasoning, and tool calling. The date label is retained for compatibility. Served only by its own checkpoint; it may be unavailable while that host is down.

DeepSeek V4.1 Flash 0910 Uncensored Thinking

deepseek/deepseek-v4.1-flash-0910-uncensored:thinking

Uncensored DeepSeek V4.1 Flash with text and image input, optional reasoning, and tool calling. The date label is retained for compatibility. Served only by its own checkpoint; it may be unavailable while that host is down.

DeepSeek V4.1 Flash 1017 Uncensored

deepseek/deepseek-v4.1-flash-1017-uncensored

Uncensored DeepSeek V4.1 Flash with text and image input, optional reasoning, and tool calling. The date label is retained for compatibility. Served only by its own checkpoint; it may be unavailable while that host is down.