What is the IQ of deepseek-v4-pro?

deepseek-v4-pro has an estimated IQ of 123, which ranks #31 out of 173 public models on the AI IQ leaderboard. The score combines seven equally weighted reasoning dimensions using measured benchmark results and conservative estimates for missing coverage.

What is deepseek-v4-pro best at?

Among the dimensions with published benchmark results, deepseek-v4-pro ranks highest in Abstract Reasoning (IQ 109, #25 of 197) and Mathematical Reasoning (IQ 132, #24 of 173). Its lowest-ranked measured dimension is Reliability (IQ 118, #54 of 173).

What is the EQ of deepseek-v4-pro?

deepseek-v4-pro has an estimated EQ of 118. EQ measures emotional reasoning — how well a model reads and responds to emotional context — and is scored separately from IQ.

Which benchmarks has deepseek-v4-pro been tested on?

deepseek-v4-pro has published results on 30 of the 39 benchmarks behind AI IQ's Composite IQ, including LiveCodeBench, GPQA Diamond, Chess Puzzles, MathArena and IFBench.

How much does deepseek-v4-pro cost?

deepseek-v4-pro costs $1.32 per million input tokens and $3.96 per million output tokens. Adjusted for how many tokens it actually consumes on real tasks, its effective cost works out to $6.95 per million input-plus-output tokens.

Is deepseek-v4-pro open source?

Yes — deepseek-v4-pro is an open-weights model: its trained weights are published, so it can be downloaded and self-hosted.

Who developed deepseek-v4-pro?

deepseek-v4-pro was developed by DeepSeek, based in China.

When was deepseek-v4-pro released?

deepseek-v4-pro was publicly released on April 24, 2026.

How large is the context window of deepseek-v4-pro?

deepseek-v4-pro supports a context window of 1,000,000 tokens — the amount of text it can consider in a single request.

How fast is deepseek-v4-pro?

In independent measurements, deepseek-v4-pro generates about 92 tokens per second, with a typical response time of about 29 seconds.

DeepSeekReleased Apr 24, 2026

Open weights

deepseek-v4-pro

Strongest in Abstract Reasoning (#25 of 197) and Mathematical Reasoning (#24 of 173); weakest in Reliability (#54 of 173). All seven dimensions are measured.

Composite IQ ?

123

#31 of 173 ranked models · ahead of 83%

DeepSeek's #3 model, behind DeepSeek V4.1 Flash 1017 (129)

Effective cost

$6.95/1M

$1.32 in · $3.96 out

Speed

92tok/s

29s typical response

Context

1Mtokens

1,000,000 tokens in

Evidence

7of 7 dims

30 of 39 benchmarks

Compare with gpt-6.1-solAPI pricing

IQ profile

Seven dimensions, averaged equally into Composite IQ

Compare shape with

Measured95% range

#1 in dimensionTop-10 median

IQ dimensions

Ranks are among models scored in each dimension. Ranges are model-based estimates, not calibrated confidence intervals.

DimensionIQPosition · 70–160RankEvidencevs Top 10

Abstract ReasoningAbstract

109measured

#25 / 197

4/5

-18.5

Mathematical ReasoningMath

132measured

#24 / 173

4/5

-14

Academic ReasoningAcademic

129measured

#40 / 199

4/6

-6

Programmatic ReasoningCoding

128measured

#42 / 174

5/6

-20.5

Computer UseComputer Use

120measured

#54 / 173

3/5

-22

ReliabilityReliability

118measured

#54 / 173

5/6

-14.5

Knowledge WorkKnowledge Work

123measured

#39 / 173

5/6

-8

deepseek-v4-pro95% range (model-based, uncalibrated)Top-10 median#1 in dimension100 = average human

Benchmark results

30 of 39 Composite IQ benchmarks have a published result for deepseek-v4-pro. Benchmarks without a result stay listed so gaps are visible.

Abstract Reasoning IQ 109 · measuredBest: #10 of 76 on Chess Puzzles4/5

BenchmarkScorePercentileRankLeader · source

Chess Puzzles

Epoch AI vEpoch 100-position leaderboard (task implementation versions recorded per run) · Oct 14

47.0%

#10 / 76

Leader gpt-6-astra · 72.0%
Epoch AI vEpoch 100-position leaderboard (task implementation versions recorded per run) · Oct 14

Mystery Game Puzzles

Full source leaderboard reconciliation vEpoch 100-position leaderboard (task implementation versions recorded per run) · Oct 23

43.0%

#13 / 55

Leader gpt-6-astra · 84.0%
Full source leaderboard reconciliation vEpoch 100-position leaderboard (task implementation versions recorded per run) · Oct 23

ARC-AGI-1

Full source leaderboard reconciliation · Oct 23

90.5%

#27 / 68

Leader fable-5 · 98.5%
Full source leaderboard reconciliation · Oct 23

ARC-AGI-2

Full source leaderboard reconciliation · Oct 23

61.3%

#28 / 68

Leader gpt-6-astra · 95.0%
Full source leaderboard reconciliation · Oct 23

Not yet reported

ARC-AGI-3

Mathematical Reasoning IQ 132 · measuredBest: #10 of 61 on MathArena4/5

BenchmarkScorePercentileRankLeader · source

MathArena

MathArena Models · Sep 22

60.8%

#10 / 61

Leader gpt-6.1-sol · 84.0%
MathArena Models · Sep 22

FrontierMath Tier 1-3

Full source leaderboard reconciliation · Oct 23

64.6%

#33 / 83

Leader gpt-6-astra · 93.7%
Full source leaderboard reconciliation · Oct 23

FrontierMath Tier 4

Full source leaderboard reconciliation · Oct 23

26.8%

#33 / 71

Leader gpt-6.1-sol · 100%
Full source leaderboard reconciliation · Oct 23

ProofBench 1.1

Full source leaderboard reconciliation v1.1 · Oct 23

50.0%

#29 / 47

Leader fable-5.1 · 100%
Full source leaderboard reconciliation v1.1 · Oct 23

Not yet reported

AIME

Academic Reasoning IQ 129 · measuredBest: #19 of 147 on GPQA Diamond4/6

BenchmarkScorePercentileRankLeader · source

GPQA Diamond

Artificial Analysis v4.3.2 · Oct 23

92.8%

#19 / 147

Leader gpt-6-astra · 96.1%
Artificial Analysis v4.3.2 · Oct 23

CritPt

Artificial Analysis v4.3.2 · Oct 23

18.0%

#30 / 157

Leader Muse Wonder Full · 32.9%
Artificial Analysis v4.3.2 · Oct 23

Humanity's Last Exam

Artificial Analysis v4.3.2 · Oct 23

41.0%

#42 / 165

Leader opus-5.5 · 61.4%
Artificial Analysis v4.3.2 · Oct 23

SciCode

Artificial Analysis v4.3.2 · Oct 23

51.0%

#57 / 165

Leader opus-5.5 · 66.9%
Artificial Analysis v4.3.2 · Oct 23

Not yet reported

MMLU-ProMMMU-Pro

Programmatic Reasoning IQ 128 · measuredBest: #6 of 61 on LiveCodeBench5/6

BenchmarkScorePercentileRankLeader · source

LiveCodeBench

Vals.ai · Aug 4

87.5%

#6 / 61

Leader Fugu Ultra v1.0 · 93.2%
Vals.ai · Aug 4

Terminal-Bench 4.0

Artificial Analysis v4.3.2 · Oct 23

14.1%

#33 / 102

Leader sonnet-5.5 · 63.6%
Artificial Analysis v4.3.2 · Oct 23

ProgramBench

Full source leaderboard reconciliation · Oct 23

13.5%

#23 / 61

Leader opus-5.5 · 65.0%
Full source leaderboard reconciliation · Oct 23

IOI V2

Full source leaderboard reconciliation v2 · Oct 23

51.6%

#24 / 42

Leader Gemini 4 Argon · 100%
Full source leaderboard reconciliation v2 · Oct 23

SWE-rebench

SWE-rebench · Sep 22

42.7%

#20 / 25

Leader fable-5 · 64.5%
SWE-rebench · Sep 22

Not yet reported

MirrorCode

Computer Use IQ 120 · measuredBest: #15 of 42 on BrowseComp3/5

BenchmarkScorePercentileRankLeader · source

BrowseComp

LLM Stats BrowseComp · Sep 22

83.4%

#15 / 42

Leader gpt-6-astra · 91.5%
LLM Stats BrowseComp · Sep 22

Arena.ai Agent Arena

Arena.ai Agent Arena · Oct 23

0.1

#37 / 64

Leader opus-5.5 · 20.9
Arena.ai Agent Arena · Oct 23

Agents' Last Exam

Agents' Last Exam · Sep 22

12.4%

#21 / 29

Leader opus-5.5 · 38.2%
Agents' Last Exam · Sep 22

Not yet reported

OSWorld-VerifiedMCP Atlas

Reliability IQ 118 · measuredBest: #17 of 95 on IFBench5/6

BenchmarkScorePercentileRankLeader · source

IFBench

Artificial Analysis model leaderboard · Aug 19

76.0%

#17 / 95

Leader minimax-m3 · 82.9%
Artificial Analysis model leaderboard · Aug 19

AA Long Context Reasoning v1.1

Artificial Analysis v4.3.2 · Oct 23

80.3%

#39 / 152

Leader kimi-k3 · 88.7%
Artificial Analysis v4.3.2 · Oct 23

SimpleQA Verified

Full source leaderboard reconciliation · Oct 23

52.9%

#26 / 79

Leader Gemini 4 Argon · 76.0%
Full source leaderboard reconciliation · Oct 23

AA Omniscience

Artificial Analysis v4.3.2 · Oct 23

0.8

#66 / 156

Leader opus-5.5 · 46.4
Artificial Analysis v4.3.2 · Oct 23

BullshitBench v2

Full source leaderboard reconciliation · Oct 23

35.0%

#46 / 83

Leader opus-4.8 · 94.0%
Full source leaderboard reconciliation · Oct 23

Not yet reported

FACTS Grounding

Knowledge Work IQ 123 · measuredBest: #4 of 19 on EnterpriseOps-Gym-AA5/6

BenchmarkScorePercentileRankLeader · source

EnterpriseOps-Gym-AA

Full source leaderboard reconciliation v1117-task-strict-2026-09-29 · Oct 23

49.6%

#4 / 19

Leader fable-5 · 51.1%
Full source leaderboard reconciliation v1117-task-strict-2026-09-29 · Oct 23

AutomationBench-AA

Artificial Analysis v4.3.2 · Oct 23

56.7%

#28 / 104

Leader Gemini 4 Argon · 77.5%
Artificial Analysis v4.3.2 · Oct 23

GDPval-AA v2.1 normalized score

Artificial Analysis v4.3.2 · Oct 23

47.4%

#36 / 124

Leader opus-5.5 · 68.3%
Artificial Analysis v4.3.2 · Oct 23

AA-Briefcase v1.1 Elo

Artificial Analysis v4.3.2 · Oct 23

1,256

#42 / 109

Leader sonnet-5.5 · 1,816
Artificial Analysis v4.3.2 · Oct 23

GDP.pdf

Artificial Analysis v4.3.2 · Oct 23

11.4%

#52 / 104

Leader Muse Wonder Full · 32.6%
Artificial Analysis v4.3.2 · Oct 23

Not yet reported

AA-AnalystAgent

Release & availability

Facts from the publisher's announcement and documentation.

Released: Apr 24, 2026
Developer: DeepSeek
Access: Available through the DeepSeek API with open weights for local deployment
Weights: Open weightsDownloadable and self-hostable
Inputs: text
Context: 1,000,000 tokens
API price: $1.32 in · $3.96 outper 1M tokens · full pricing

What DeepSeek says

The 0813 checkpoint is tracked as a dated revision of the existing DeepSeek V4 Pro identity.
The canonical model row retains the original April launch date while current benchmark data follows the latest revision.

Limitations

DeepSeek uses a mutable canonical API alias, so benchmark results can change without a new public model ID.

Publisher statements, not AI IQ measurements.

Sourcesapi-docs.deepseek.com/updates vals.ai/models/deepseek_v4_pro_0813artificialanalysis.ai/models/deepseek-v4-pro

About this score

How much of deepseek-v4-pro's IQ is measured, and how the rest is estimated.

30/ 39benchmark results

7/ 7dimensions measured

0estimated

4/5Abstract

4/5Math

4/6Academic

5/6Coding

3/5Computer Use

5/6Reliability

5/6Knowledge Work

Bar height = share of benchmarks with resultsEstimatedNo score

Measured results are shown as published

Every score in the benchmark tables comes from its source unchanged. Composite IQ averages all seven dimensions equally.

How gaps are handled

Every dimension has at least one published result.

A missing result is not a zero

Benchmark versions and evaluation harnesses can differ between sources, so small gaps between models may not be meaningful.

Read the scoring methodology →

Compare deepseek-v4-pro

Side-by-side pages for the models it is most often weighed against.

#1 overall

gpt-6.1-sol

IQ141 18 higherEffective cost$6.25 1.1× cheaper

DeepSeek #1

DeepSeek V4.1 Flash 1017

IQ129 6 higherEffective cost$2.40 2.9× cheaper

Closest IQ

MiniMax M3.1 Flash

IQ123 same IQEffective cost$1.40 5.0× cheaper

Most similar profile

gpt-5.4

IQ122 1 lowerEffective cost$20.47 2.9× pricier

Similar price

qwen3.7-max

IQ118 5 lowerEffective cost$7.44 about the same

#2 overall

opus-5.5

IQ141 18 higherEffective cost$35.41 5.1× pricier

deepseek-v4-proComparison model

Compare with any modelBrowse all comparisons →

Frequently asked questions

Quick answers about deepseek-v4-pro, drawn from the same data as the rest of this page.

How AI IQ scores models →Get this data from the API →

What is the IQ of deepseek-v4-pro?: deepseek-v4-pro has an estimated IQ of 123, which ranks #31 out of 173 public models on the AI IQ leaderboard. The score combines seven equally weighted reasoning dimensions using measured benchmark results and conservative estimates for missing coverage.
What is deepseek-v4-pro best at?: Among the dimensions with published benchmark results, deepseek-v4-pro ranks highest in Abstract Reasoning (IQ 109, #25 of 197) and Mathematical Reasoning (IQ 132, #24 of 173). Its lowest-ranked measured dimension is Reliability (IQ 118, #54 of 173).
What is the EQ of deepseek-v4-pro?: deepseek-v4-pro has an estimated EQ of 118. EQ measures emotional reasoning — how well a model reads and responds to emotional context — and is scored separately from IQ.
Which benchmarks has deepseek-v4-pro been tested on?: deepseek-v4-pro has published results on 30 of the 39 benchmarks behind AI IQ's Composite IQ, including LiveCodeBench, GPQA Diamond, Chess Puzzles, MathArena and IFBench.
How much does deepseek-v4-pro cost?: deepseek-v4-pro costs $1.32 per million input tokens and $3.96 per million output tokens. Adjusted for how many tokens it actually consumes on real tasks, its effective cost works out to $6.95 per million input-plus-output tokens.
Is deepseek-v4-pro open source?: Yes — deepseek-v4-pro is an open-weights model: its trained weights are published, so it can be downloaded and self-hosted.
Who developed deepseek-v4-pro?: deepseek-v4-pro was developed by DeepSeek, based in China.
When was deepseek-v4-pro released?: deepseek-v4-pro was publicly released on April 24, 2026.
How large is the context window of deepseek-v4-pro?: deepseek-v4-pro supports a context window of 1,000,000 tokens — the amount of text it can consider in a single request.
How fast is deepseek-v4-pro?: In independent measurements, deepseek-v4-pro generates about 92 tokens per second, with a typical response time of about 29 seconds.