Provider logo

DeepSeek V4 Pro

deepseek/deepseek-v4-pro
Provider logo

DeepSeek V4 Pro

deepseek/deepseek-v4-pro

DeepSeek V4 Pro is a large-scale Mixture-of-Experts model from DeepSeek with a 1M-token context window, built for advanced reasoning, coding, long-horizon agent workflows, knowledge, math, and software engineering tasks.

Added Apr 24, 2026

Model weights

Context Window

1.0M

Max Output

384.0K

Avg output tokens (7d)

471 tokens

40%

Input Price (Auto)

$1.10/1M

Output Price (Auto)

$2.20/1M

Cache Read (Auto)

$0.11/1M

Capabilities

Benchmarks

Performance metrics and benchmarks

Sourced from Artificial Analysis.

Intelligence Index

36.3

Better than 91% of models compared

Coding Index

68.8

Better than 80% of models compared

Agentic work

AutomationBench-AA

Workflow automation with guardrail penalties

56.7%

Better than 84% of models compared

AutomationBench-AA Tasks Completed

Fully completed workflows without guardrail violations

25.9%

Better than 46% of models compared

AA-Briefcase

Agentic knowledge work (Elo)

1265 Elo

Better than 73% of models compared

GDPval-AA v2

Economically valuable tasks (Elo)

1493 Elo

Better than 86% of models compared

Document reasoning

GDP.pdf

Professional PDF reasoning: all-pass rate

11.4%

Better than 53% of models compared

AA-LCR v1.1

Long context reasoning with updated grading

80.3%

Better than 91% of models compared

Reasoning

HLE

Humanity's Last Exam

41.0%

Better than 92% of models compared

IFBench

Instruction-following benchmark

45.8%

Better than 53% of models compared

CritPt

Research-level physics reasoning

0.9%

Coding

Terminal-Bench v4.0

Practical coding and terminal tasks

14.1%

Better than 75% of models compared

SciCode

Python programming for scientific computing

51.0%

Better than 63% of models compared

Knowledge

AA-Omniscience Accuracy

Proportion of correctly answered questions

30.9%

AA-Omniscience Hallucination Rate

Rate of incorrect answers among non-correct responses

87.8%

Legacy benchmarks

GPQA Diamond (legacy)

Graduate-level scientific reasoning

92.8%

Better than 95% of models compared

Terminal-Bench Hard (legacy)

Agentic coding and terminal use

36.4%

Better than 84% of models compared

T²-Bench Telecom (legacy)

Conversational AI agents in dual-control scenarios

91.2%

Better than 86% of models compared

AA-LCR (unversioned / legacy)

Long context reasoning evaluation

80.3%

Better than 91% of models compared

Last updated Sep 9, 2026

Artificial Analysis

Providers

Choose explicit providers for this model. Auto routing remains available as the default option.

Loading provider options…

Compare DeepSeek V4 Pro with similar models from the same provider or model family.

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 Pro (Thinking)

deepseek/deepseek-v4-pro:thinking

DeepSeek V4 Pro Thinking enables DeepSeek's chain-of-thought mode on the large-scale Mixture-of-Experts model with a 1M-token context window, built for advanced reasoning, coding, long-horizon agent workflows, knowledge, math, and software engineering tasks.

DeepSeek V4.1 Flash

deepseek/deepseek-v4.1-flash

DeepSeek V4.1 Flash supports text and image input, reasoning, tool calling, and structured output with a 1M-token context window. This is a rate-limited beta with limited capacity, intended for testing rather than production use. Assume prompts and responses are logged by the provider and may be used for model training or service improvement. Do not send sensitive or confidential data.

DeepSeek V4 Flash Vision Exp Uncensored

deepseek/deepseek-v4-flash-vision-exp-uncensored

An uncensored variant of the experimental vision-enabled DeepSeek V4 Flash model for chat, image understanding, reasoning, coding, and tool use, with a 524K context window.

DeepSeek V4 Flash Vision Exp

deepseek/deepseek-v4-flash-vision-exp

An experimental vision-enabled DeepSeek V4 Flash model that adds image understanding while retaining the text, reasoning, coding, tool-calling, and agent capabilities of the base model.