OpenAI
Released September 3, 2026Cutoff April 2026

GPT-6 Astra

OpenAI's GPT-6 Astra: first model at the Critical cybersecurity threshold, computer use, 1.05M context, reasoning up to max.

Inputs
Text
Image
Outputs
Text

Model Overview

Capabilities, design details, and architectural traits

GPT-6 Astra - OpenAI's flagship model

GPT-6 Astra is OpenAI's flagship model for complex reasoning and coding. It is the first model OpenAI has designated at the Critical cybersecurity capability threshold under its Preparedness Framework, meaning that with the right tools and access it can find previously unknown security flaws and develop exploits for them without a person guiding each step.

What sets GPT-6 Astra apart

TraitDetail
First Critical-cyber designationFirst OpenAI model to meet the Critical cybersecurity capability threshold under the Preparedness Framework, requiring stronger safeguards during development and before release
Zero-day discoveryDuring evaluation it discovered and used two zero-day V8 vulnerabilities as part of an exploit chain, which OpenAI is disclosing to maintainers
Reasoning without an off switchReasoning effort runs from low to max with no "none" setting, unlike the GPT-5.6 family
Scale1.05M token context window and 128K token max output
Built-in toolsFunctions, Web search, File search, and Computer use

Professional work and efficiency

Astra pairs computer-use advances with targeted training for professional environments. It carries out multistep workflows and produces polished documents, spreadsheets, and presentations.

Guarded cyber access

Because of its critical cyber capabilities, Astra's most advanced cybersecurity work is not in the default production configuration. It is initially limited to a group of testers, with defensive access expanding through Daybreak Blue.

Benchmark Performance

Independent evaluations · Artificial Analysis

61.2%
Intelligence
76.9%
Coding Index
51.5%
Agentic Index

Accuracy & Capability Details

GPQA - Graduate Science96.1%
Humanity's Last Exam54.7%
SciCode - Scientific Coding54.1%
Long Context Reasoning74.3%

Compare Models Side-by-Side

Evaluate specifications, pricing, and independent benchmark indices

Model Details
General Info
ProviderOpenAIAnthropicAnthropic
Release DateSeptember 3, 2026September 1, 2026July 24, 2026
Knowledge CutoffApr 2026-May 2026
Context & Limits
Context Window
1.1M
Best Context Window
-1M
Pricing (per 1M tokens)
Input Pricing$10$10
$5
Best Input Pricing
Output Pricing$50$50
$25
Best Output Pricing
Modalities
Inputs
textimage
textimagefile
textimage
Outputs
text
text
text
Benchmarks (0-100)
Intelligence Index61.2
65.7
Best Intelligence Index
63.1
Coding Index76.9
81.6
Best Coding Index
78.0
Agentic Index51.5
61.3
Best Agentic Index
59.2
GPT-6 Astra
Claude Fable 5.1
Claude Opus 5

GPQA Benchmark

Graduate-level reasoning and expert Q&A evaluation.

96%
GPT-6 Astra
GPQA Benchmark
Score: 96%
GPT-6 Astra
94%
Claude Fable 5.1
GPQA Benchmark
Score: 94%
Claude Fable 5.1
93%
Claude Opus 5
GPQA Benchmark
Score: 93%
Claude Opus 5

Humanity's Last Exam

Extremely difficult logical reasoning and knowledge.

55%
GPT-6 Astra
Humanity's Last Exam
Score: 55%
GPT-6 Astra
59%
Claude Fable 5.1
Humanity's Last Exam
Score: 59%
Claude Fable 5.1
55%
Claude Opus 5
Humanity's Last Exam
Score: 55%
Claude Opus 5

Long Context Reasoning

Logical reasoning over long context windows.

74%
GPT-6 Astra
Long Context Reasoning
Score: 74%
GPT-6 Astra
80%
Claude Fable 5.1
Long Context Reasoning
Score: 80%
Claude Fable 5.1
76%
Claude Opus 5
Long Context Reasoning
Score: 76%
Claude Opus 5

Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.