What is the IQ of gpt-6-astra?

gpt-6-astra has an estimated IQ of 140, which ranks #4 out of 170 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 gpt-6-astra best at?

Among the dimensions with published benchmark results, gpt-6-astra ranks highest in Abstract Reasoning (IQ 136, #1 of 193) and Reliability (IQ 136, #1 of 170). Its lowest-ranked measured dimension is Academic Reasoning (IQ 134, #10 of 195).

What is the EQ of gpt-6-astra?

gpt-6-astra doesn't have an EQ score yet. EQ (emotional reasoning) is calculated from EQ-Bench 4 and AttuneBench, and neither has published a result for gpt-6-astra. AI IQ doesn't estimate EQ from other benchmarks, so the score appears once one of those results is published.

Which benchmarks has gpt-6-astra been tested on?

gpt-6-astra has published results on 31 of the 39 benchmarks behind AI IQ's Composite IQ, including GPQA Diamond, FrontierMath Tier 1-3, Chess Puzzles, ARC-AGI-2 and Mystery Game Puzzles.

How much does gpt-6-astra cost?

gpt-6-astra costs $10.00 per million input tokens and $50.00 per million output tokens. Adjusted for how many tokens it actually consumes on real tasks, its effective cost works out to $34.14 per million input-plus-output tokens.

Is gpt-6-astra open source?

No — gpt-6-astra is a proprietary model. Its weights are not publicly available, so it is used through hosted APIs and products.

Who developed gpt-6-astra?

gpt-6-astra was developed by OpenAI, based in the United States.

When was gpt-6-astra released?

gpt-6-astra was publicly released on September 3, 2026.

How large is the context window of gpt-6-astra?

gpt-6-astra supports a context window of 1,050,000 tokens — the amount of text it can consider in a single request.

How fast is gpt-6-astra?

In independent measurements, gpt-6-astra generates about 69 tokens per second, with a typical response time of about 300 seconds.

OpenAIReleased Sep 3, 2026

Proprietary

gpt-6-astra

Strongest in Abstract Reasoning (#1 of 193) and Reliability (#1 of 170); weakest in Academic Reasoning (#10 of 195). All seven dimensions are measured.

Composite IQ ?

140

#4 of 170 ranked models · ahead of 98%

OpenAI's #2 model, behind gpt-6.1-sol (141)

Effective cost

$34.14/1M

$10 in · $50 out

Speed

69tok/s

300s typical response

Context

1Mtokens

1,050,000 tokens in

Evidence

7of 7 dims

31 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

136measured

#1 / 193

5/5

+7.5

Mathematical ReasoningMath

149measured

#2 / 170

4/5

+2.5

Academic ReasoningAcademic

134measured

#10 / 195

5/6

-1.5

Programmatic ReasoningCoding

151measured

#4 / 171

4/6

Computer UseComputer Use

140measured

#7 / 170

3/5

-0.5

ReliabilityReliability

136measured

#1 / 170

5/6

+3.5

Knowledge WorkKnowledge Work

133measured

#6 / 170

5/6

gpt-6-astra95% range (model-based, uncalibrated)Top-10 median#1 in dimension100 = average human

Benchmark results

31 of 39 Composite IQ benchmarks have a published result for gpt-6-astra. Benchmarks without a result stay listed so gaps are visible.

Abstract Reasoning IQ 136 · measuredBest: #1 of 76 on Chess Puzzles5/5

BenchmarkScorePercentileRankLeader · source

Chess Puzzles

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

72.0%

#1 / 76

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

ARC-AGI-2

Full source leaderboard reconciliation · Oct 19

95.0%

#1 / 63

Leader on this benchmark
Full source leaderboard reconciliation · Oct 19

Mystery Game Puzzles

Full source leaderboard reconciliation vEpoch 100-position leaderboard (task implementation versions recorded per run) · Oct 19

84.0%

#1 / 55

Leader on this benchmark
Full source leaderboard reconciliation vEpoch 100-position leaderboard (task implementation versions recorded per run) · Oct 19

ARC-AGI-3

Full source leaderboard reconciliation · Oct 19

62.7%

#1 / 22

Leader on this benchmark
Full source leaderboard reconciliation · Oct 19

ARC-AGI-1

Full source leaderboard reconciliation · Oct 19

98.5%

#4 / 63

Leader fable-5 · 98.5%
Full source leaderboard reconciliation · Oct 19

Mathematical Reasoning IQ 149 · measuredBest: #1 of 82 on FrontierMath Tier 1-34/5

BenchmarkScorePercentileRankLeader · source

FrontierMath Tier 1-3

Full source leaderboard reconciliation · Oct 19

93.7%

#1 / 82

Leader on this benchmark
Full source leaderboard reconciliation · Oct 19

FrontierMath Tier 4

Full source leaderboard reconciliation · Oct 19

97.6%

#2 / 70

Leader gpt-6.1-sol · 100%
Full source leaderboard reconciliation · Oct 19

MathArena

Full source leaderboard reconciliation · Oct 19

81.9%

#2 / 61

Leader gpt-6.1-sol · 84.0%
Full source leaderboard reconciliation · Oct 19

ProofBench 1.1

Full source leaderboard reconciliation v1.1 · Oct 19

99.0%

#5 / 46

Leader fable-5.1 · 100%
Full source leaderboard reconciliation v1.1 · Oct 19

Not yet reported

AIME

Academic Reasoning IQ 134 · measuredBest: #1 of 147 on GPQA Diamond5/6

BenchmarkScorePercentileRankLeader · source

GPQA Diamond

Artificial Analysis v4.3.2 · Oct 19

96.1%

#1 / 147

Leader on this benchmark
Artificial Analysis v4.3.2 · Oct 19

CritPt

Artificial Analysis v4.3.2 · Oct 19

31.7%

#3 / 155

Leader Muse Wonder Full · 32.9%
Artificial Analysis v4.3.2 · Oct 19

Humanity's Last Exam

Artificial Analysis v4.3.2 · Oct 19

54.7%

#7 / 163

Leader opus-5.5 · 61.4%
Artificial Analysis v4.3.2 · Oct 19

MMMU-Pro

Artificial Analysis v4.3.2 · Oct 19

86.9%

#4 / 86

Leader fable-5.1 · 90.6%
Artificial Analysis v4.3.2 · Oct 19

SciCode

Artificial Analysis v4.3.2 · Oct 19

56.5%

#24 / 163

Leader opus-5.5 · 66.9%
Artificial Analysis v4.3.2 · Oct 19

Not yet reported

MMLU-Pro

Programmatic Reasoning IQ 151 · measuredBest: #3 of 100 on Terminal-Bench 4.04/6

BenchmarkScorePercentileRankLeader · source

Terminal-Bench 4.0

Artificial Analysis v4.3.2 · Oct 19

59.1%

#3 / 100

Leader sonnet-5.5 · 63.6%
Artificial Analysis v4.3.2 · Oct 19

IOI V2

Full source leaderboard reconciliation v2 · Oct 19

100%

#2 / 41

Leader Gemini 4 Argon · 100%
Full source leaderboard reconciliation v2 · Oct 19

ProgramBench

Full source leaderboard reconciliation · Oct 19

50.0%

#3 / 61

Leader opus-5.5 · 65.0%
Full source leaderboard reconciliation · Oct 19

MirrorCode

Full source leaderboard reconciliation vML,+Private,2L · Oct 19

46.7%

#4 / 9

Leader opus-5.5 · 77.4%
Full source leaderboard reconciliation vML,+Private,2L · Oct 19

Not yet reported

LiveCodeBenchSWE-rebench

Computer Use IQ 140 · measuredBest: #1 of 39 on BrowseComp3/5

BenchmarkScorePercentileRankLeader · source

BrowseComp

Full source leaderboard reconciliation · Oct 19

91.5%

#1 / 39

Leader on this benchmark
Full source leaderboard reconciliation · Oct 19

Agents' Last Exam

Full source leaderboard reconciliation · Oct 19

34.2%

#2 / 29

Leader opus-5.5 · 38.2%
Full source leaderboard reconciliation · Oct 19

Arena.ai Agent Arena

Arena.ai Agent Arena · Oct 13

12.3

#5 / 59

Leader opus-5.5 · 17.9
Arena.ai Agent Arena · Oct 13

Not yet reported

OSWorld-VerifiedMCP Atlas

Reliability IQ 136 · measuredBest: #3 of 154 on AA Omniscience5/6

BenchmarkScorePercentileRankLeader · source

AA Omniscience

Artificial Analysis v4.3.2 · Oct 19

43.4

#3 / 154

Leader opus-5.5 · 46.4
Artificial Analysis v4.3.2 · Oct 19

SimpleQA Verified

Full source leaderboard reconciliation · Oct 19

75.6%

#2 / 78

Leader Gemini 4 Argon · 76.0%
Full source leaderboard reconciliation · Oct 19

FACTS Grounding

Full source leaderboard reconciliation · Oct 19

82.0%

#3 / 36

Leader gpt-6-sol · 85.2%
Full source leaderboard reconciliation · Oct 19

BullshitBench v2

Full source leaderboard reconciliation · Oct 19

69.0%

#11 / 83

Leader opus-4.8 · 94.0%
Full source leaderboard reconciliation · Oct 19

AA Long Context Reasoning v1.1

Artificial Analysis v4.3.2 · Oct 19

80.7%

#35 / 150

Leader kimi-k3 · 88.7%
Artificial Analysis v4.3.2 · Oct 19

Not yet reported

IFBench

Knowledge Work IQ 133 · measuredBest: #2 of 102 on GDP.pdf5/6

BenchmarkScorePercentileRankLeader · source

GDP.pdf

Artificial Analysis v4.3.2 · Oct 19

31.0%

#2 / 102

Leader Muse Wonder Full · 32.6%
Artificial Analysis v4.3.2 · Oct 19

AutomationBench-AA

Artificial Analysis v4.3.2 · Oct 19

68.5%

#6 / 102

Leader Gemini 4 Argon · 77.5%
Artificial Analysis v4.3.2 · Oct 19

AA-Briefcase v1.1 Elo

Artificial Analysis v4.3.2 · Oct 19

1,569

#13 / 107

Leader sonnet-5.5 · 1,817
Artificial Analysis v4.3.2 · Oct 19

GDPval-AA v2.1 normalized score

Artificial Analysis v4.3.2 · Oct 19

53.6%

#25 / 121

Leader opus-5.5 · 68.3%
Artificial Analysis v4.3.2 · Oct 19

AA-AnalystAgent

Full source leaderboard reconciliation vpass^5-2026-09-29 · Oct 19

51.3%

#8 / 30

Leader gemini-3.7-flash · 60.0%
Full source leaderboard reconciliation vpass^5-2026-09-29 · Oct 19

Not yet reported

EnterpriseOps-Gym-AA

Release & availability

Facts from the publisher's announcement and documentation.

Released: Sep 3, 2026
Developer: OpenAI
Access: Rolling out to ChatGPT Plus, Pro, Business, and Enterprise, and to developers through the OpenAI API, Microsoft Azure, and AWS Bedrock
Weights: ProprietaryHosted access only
Inputs: textimage
Context: 1,050,000 tokens
API price: $10 in · $50 outper 1M tokens · full pricing

What OpenAI says

OpenAI positions GPT-6 Astra as its most capable model for complex reasoning, coding, computer use, research, and document creation.
The API supports low, medium, high, xhigh, and max reasoning effort with a 1.05M-token context window.

Limitations

Access is rolling out in phases rather than being immediately available to every eligible user.
OpenAI reports reduced chain-of-thought monitorability relative to GPT-5.6 Sol and applies additional monitoring and access controls because Astra reaches its Critical cybersecurity capability threshold.

Publisher statements, not AI IQ measurements.

Sourcesopenai.com/index/gpt-6-astra developers.openai.com/api/docs/models/gpt-6-astraopenai.com/index/safety-overview-gpt-6-astradeploymentsafety.openai.com/gpt-6-astra/vision

About this score

How much of gpt-6-astra's IQ is measured, and how the rest is estimated.

31/ 39benchmark results

7/ 7dimensions measured

0estimated

5/5Abstract

4/5Math

5/6Academic

4/6Coding

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. Missing benchmarks can also draw on its predecessors, GPT 5.6 Sol and GPT 5.5; a predecessor is an estimation input, not evidence that this model has the predecessor's measured results.

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 gpt-6-astra

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

#1 overall

gpt-6.1-sol

IQ141 1 higherEffective cost$6.29 5.4× cheaper

Predecessor

gpt-5.6-sol

IQ134 6 lowerEffective cost$27.43 1.2× cheaper

IQ141 1 higherEffective cost$35.67 about the same

Similar price

Muse Wonder

IQ130 10 lowerEffective cost$31.40 about the same

#5 overall

fable-5.1

IQ137 3 lowerEffective cost$90.82 2.7× pricier

gpt-6-astraComparison model

Compare with any modelBrowse all comparisons →

Frequently asked questions

Quick answers about gpt-6-astra, 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 gpt-6-astra?: gpt-6-astra has an estimated IQ of 140, which ranks #4 out of 170 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 gpt-6-astra best at?: Among the dimensions with published benchmark results, gpt-6-astra ranks highest in Abstract Reasoning (IQ 136, #1 of 193) and Reliability (IQ 136, #1 of 170). Its lowest-ranked measured dimension is Academic Reasoning (IQ 134, #10 of 195).
What is the EQ of gpt-6-astra?: gpt-6-astra doesn't have an EQ score yet. EQ (emotional reasoning) is calculated from EQ-Bench 4 and AttuneBench, and neither has published a result for gpt-6-astra. AI IQ doesn't estimate EQ from other benchmarks, so the score appears once one of those results is published.
Which benchmarks has gpt-6-astra been tested on?: gpt-6-astra has published results on 31 of the 39 benchmarks behind AI IQ's Composite IQ, including GPQA Diamond, FrontierMath Tier 1-3, Chess Puzzles, ARC-AGI-2 and Mystery Game Puzzles.
How much does gpt-6-astra cost?: gpt-6-astra costs $10.00 per million input tokens and $50.00 per million output tokens. Adjusted for how many tokens it actually consumes on real tasks, its effective cost works out to $34.14 per million input-plus-output tokens.
Is gpt-6-astra open source?: No — gpt-6-astra is a proprietary model. Its weights are not publicly available, so it is used through hosted APIs and products.
Who developed gpt-6-astra?: gpt-6-astra was developed by OpenAI, based in the United States.
When was gpt-6-astra released?: gpt-6-astra was publicly released on September 3, 2026.
How large is the context window of gpt-6-astra?: gpt-6-astra supports a context window of 1,050,000 tokens — the amount of text it can consider in a single request.
How fast is gpt-6-astra?: In independent measurements, gpt-6-astra generates about 69 tokens per second, with a typical response time of about 300 seconds.