What is the IQ of deepseek-r1?

deepseek-r1 has an estimated IQ of 93, which ranks #154 out of 176 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-r1 best at?

Among the dimensions with published benchmark results, deepseek-r1 ranks highest in Programmatic Reasoning (IQ 104, #113 of 177) and Academic Reasoning (IQ 108, #141 of 203). Its lowest-ranked measured dimension is Abstract Reasoning (IQ 72, #194 of 201).

What is the EQ of deepseek-r1?

deepseek-r1 has an estimated EQ of 100. EQ measures emotional reasoning — how well a model reads and responds to emotional context — and is scored separately from IQ.

Which benchmarks has deepseek-r1 been tested on?

deepseek-r1 has published results on 19 of the 39 benchmarks behind AI IQ's Composite IQ, including MMLU-Pro, Terminal-Bench 4.0, CritPt, LiveCodeBench and SciCode.

How much does deepseek-r1 cost?

deepseek-r1 costs $2.00 per million input tokens and $4.00 per million output tokens. Adjusted for how many tokens it actually consumes on real tasks, its effective cost works out to $3.12 per million input-plus-output tokens.

Is deepseek-r1 open source?

Yes — deepseek-r1 is an open-weights model: its trained weights are published, so it can be downloaded and self-hosted.

Who developed deepseek-r1?

deepseek-r1 was developed by DeepSeek, based in China.

When was deepseek-r1 released?

deepseek-r1 was publicly released on January 20, 2025.

How large is the context window of deepseek-r1?

AI IQ hasn't verified an input context limit for deepseek-r1 from its developer's documentation, so none is listed rather than guessing.

How fast is deepseek-r1?

In independent measurements, deepseek-r1 generates about 165 tokens per second, with a typical response time of about 18 seconds.

DeepSeekReleased Jan 20, 2025

Open weights

deepseek-r1

Its best results are in Programmatic Reasoning (#113 of 177) and Academic Reasoning (#141 of 203); weakest in Abstract Reasoning (#194 of 201). Computer Use is estimated, because no results are published yet.

Composite IQ ?

#154 of 176 ranked models · ahead of 13%

DeepSeek's #8 model, behind DeepSeek V4.1 Flash 1017 (128)

Effective cost

$3.12/1M

$2 in · $4 out

Speed

165tok/s

18s typical response

Context

Unverified

Not in publisher docs

Evidence

6of 7 dims

19 of 39 benchmarks

Compare with gpt-6.1-solAPI pricing

IQ profile

Seven dimensions, averaged equally into Composite IQ

Compare shape with

MeasuredEstimated95% 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

72measured

#194 / 201

2/5

-54.5

Mathematical ReasoningMath

101measured

#124 / 176

2/5

-46

Academic ReasoningAcademic

108measured

#141 / 203

5/6

-27.5

Programmatic ReasoningCoding

104measured

#113 / 177

2/6

-45.5

Computer UseComputer Use

70estimated

#167 / 176

0/5

-73

ReliabilityReliability

98measured

#156 / 177

4/6

-36

Knowledge WorkKnowledge Work

96measured

#130 / 176

4/6

-35.5

deepseek-r195% range (model-based, uncalibrated)Top-10 median#1 in dimension100 = average human

Benchmark results

19 of 39 Composite IQ benchmarks have a published result for deepseek-r1. Benchmarks without a result stay listed so gaps are visible.

Abstract Reasoning IQ 72 · measuredBest: #62 of 68 on ARC-AGI-22/5

BenchmarkScorePercentileRankLeader · source

ARC-AGI-2

ARC Prize verified leaderboard · Sep 22

1.3%

#62 / 68

Leader gpt-6-astra · 95.0%
ARC Prize verified leaderboard · Sep 22

ARC-AGI-1

ARC Prize verified leaderboard · Sep 22

15.8%

#63 / 68

Leader fable-5 · 98.5%
ARC Prize verified leaderboard · Sep 22

Not yet reported

ARC-AGI-3Chess PuzzlesMystery Game Puzzles

Mathematical Reasoning IQ 101 · measuredBest: #44 of 56 on AIME2/5

BenchmarkScorePercentileRankLeader · source

AIME

74.0%

#44 / 56

Leader gemini-3.1-pro · 98.1%

MathArena

Full source leaderboard reconciliation · Oct 28

8.8%

#61 / 64

Leader gpt-6.1-sol · 84.0%
Full source leaderboard reconciliation · Oct 28

Not yet reported

FrontierMath Tier 4FrontierMath Tier 1-3ProofBench 1.1

Academic Reasoning IQ 108 · measuredBest: #25 of 45 on MMLU-Pro5/6

BenchmarkScorePercentileRankLeader · source

MMLU-Pro

TIGER-Lab MMLU-Pro leaderboard (missing rows only) · Sep 22

83.2%

#25 / 45

Leader fable-5.1 · 92.4%
TIGER-Lab MMLU-Pro leaderboard (missing rows only) · Sep 22

CritPt

Artificial Analysis v4.3.2 · Oct 27

0.6%

#122 / 159

Leader Muse Wonder Full · 32.9%
Artificial Analysis v4.3.2 · Oct 27

SciCode

Artificial Analysis v4.3.2 · Oct 27

38.3%

#130 / 167

Leader opus-5.5 · 66.9%
Artificial Analysis v4.3.2 · Oct 27

GPQA Diamond

Artificial Analysis v4.3.2 · Oct 27

70.8%

#126 / 147

Leader gpt-6-astra · 96.1%
Artificial Analysis v4.3.2 · Oct 27

Humanity's Last Exam

Artificial Analysis v4.3.2 · Oct 27

8.5%

#150 / 167

Leader opus-5.5 · 61.4%
Artificial Analysis v4.3.2 · Oct 27

Not yet reported

MMMU-Pro

Programmatic Reasoning IQ 104 · measuredBest: #74 of 104 on Terminal-Bench 4.02/6

BenchmarkScorePercentileRankLeader · source

Terminal-Bench 4.0

Artificial Analysis v4.3.2 · Oct 27

0.0%

#74 / 104

Leader Kimi K3.1 Pro · 67.2%
Artificial Analysis v4.3.2 · Oct 27

LiveCodeBench

Vals.ai · Aug 4

70.2%

#47 / 61

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

Not yet reported

IOI V2ProgramBenchSWE-rebenchMirrorCode

Computer Use IQ 70 · estimatedNo published results0/5

No published results yet for Computer Use. The 70 shown is estimated from deepseek-r1's other dimensions and will be replaced as results come in. How estimates work

BrowseCompOSWorld-VerifiedMCP AtlasArena.ai Agent ArenaAgents' Last Exam

Reliability IQ 98 · measuredBest: #126 of 158 on AA Omniscience4/6

BenchmarkScorePercentileRankLeader · source

AA Omniscience

Artificial Analysis v4.3.2 · Oct 27

-32.2

#126 / 158

Leader opus-5.5 · 46.4
Artificial Analysis v4.3.2 · Oct 27

AA Long Context Reasoning v1.1

Artificial Analysis v4.3.2 · Oct 27

57.7%

#132 / 154

Leader kimi-k3 · 88.7%
Artificial Analysis v4.3.2 · Oct 27

BullshitBench v2

Full source leaderboard reconciliation · Oct 23

8.0%

#77 / 83

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

IFBench

Artificial Analysis v4.3.2 · Oct 27

39.0%

#90 / 96

Leader minimax-m3 · 82.9%
Artificial Analysis v4.3.2 · Oct 27

Not yet reported

SimpleQA VerifiedFACTS Grounding

Knowledge Work IQ 96 · measuredBest: #103 of 124 on GDPval-AA v2.1 normalized score4/6

BenchmarkScorePercentileRankLeader · source

GDPval-AA v2.1 normalized score

Artificial Analysis v4.3.2 · Oct 20

0.0%

#103 / 124

Leader opus-5.5 · 68.3%
Artificial Analysis v4.3.2 · Oct 20

GDP.pdf

Artificial Analysis v4.3.2 · Oct 27

2.4%

#95 / 106

Leader Muse Wonder Full · 32.6%
Artificial Analysis v4.3.2 · Oct 27

AA-Briefcase v1.1 Elo

Artificial Analysis v4.3.2 · Oct 27

128.1

#104 / 111

Leader sonnet-5.5 · 1,812
Artificial Analysis v4.3.2 · Oct 27

AutomationBench-AA

Artificial Analysis v4.3.2 · Oct 27

0.2%

#104 / 106

Leader Gemini 4 Argon · 77.5%
Artificial Analysis v4.3.2 · Oct 27

Not yet reported

AA-AnalystAgentEnterpriseOps-Gym-AA

Release & availability

Release details on record for this model.

Released: Jan 20, 2025
Developer: DeepSeek
Weights: Open weightsDownloadable and self-hostable
API price: $2 in · $4 outper 1M tokens · full pricing

About this score

How much of deepseek-r1's IQ is measured, and how the rest is estimated.

19/ 39benchmark results

6/ 7dimensions measured

1estimated

2/5Abstract

2/5Math

5/6Academic

2/6Coding

0/5Computer Use

4/6Reliability

4/6Knowledge 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.

Estimates fill the gaps

Computer Use has no published results yet, so its IQ is estimated from deepseek-r1's other dimensions.

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-r1

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

#1 overall

gpt-6.1-sol

IQ141 48 higherEffective cost$6.30 2.0× pricier

DeepSeek #1

DeepSeek V4.1 Flash 1017

IQ128 35 higherEffective cost$2.42 1.3× cheaper

Closest IQ

trinity-large

IQ93 same IQEffective cost$1.04 3.0× cheaper

Most similar profile

kimi-k2

IQ91 2 lowerEffective cost$2.38 1.3× cheaper

Similar price

haiku-4.5

IQ101 8 higherEffective cost$3.00 about the same

#2 overall

opus-5.5

IQ141 48 higherEffective cost$35.73 11.4× pricier

deepseek-r1Comparison model

Compare with any modelBrowse all comparisons →

Frequently asked questions

Quick answers about deepseek-r1, 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-r1?: deepseek-r1 has an estimated IQ of 93, which ranks #154 out of 176 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-r1 best at?: Among the dimensions with published benchmark results, deepseek-r1 ranks highest in Programmatic Reasoning (IQ 104, #113 of 177) and Academic Reasoning (IQ 108, #141 of 203). Its lowest-ranked measured dimension is Abstract Reasoning (IQ 72, #194 of 201).
What is the EQ of deepseek-r1?: deepseek-r1 has an estimated EQ of 100. EQ measures emotional reasoning — how well a model reads and responds to emotional context — and is scored separately from IQ.
Which benchmarks has deepseek-r1 been tested on?: deepseek-r1 has published results on 19 of the 39 benchmarks behind AI IQ's Composite IQ, including MMLU-Pro, Terminal-Bench 4.0, CritPt, LiveCodeBench and SciCode.
How much does deepseek-r1 cost?: deepseek-r1 costs $2.00 per million input tokens and $4.00 per million output tokens. Adjusted for how many tokens it actually consumes on real tasks, its effective cost works out to $3.12 per million input-plus-output tokens.
Is deepseek-r1 open source?: Yes — deepseek-r1 is an open-weights model: its trained weights are published, so it can be downloaded and self-hosted.
Who developed deepseek-r1?: deepseek-r1 was developed by DeepSeek, based in China.
When was deepseek-r1 released?: deepseek-r1 was publicly released on January 20, 2025.
How large is the context window of deepseek-r1?: AI IQ hasn't verified an input context limit for deepseek-r1 from its developer's documentation, so none is listed rather than guessing.
How fast is deepseek-r1?: In independent measurements, deepseek-r1 generates about 165 tokens per second, with a typical response time of about 18 seconds.