Updated

A floating star-shaped lattice of tiny suspended glass beads joined by hair-thin silver threads on a dark background, abstract hero artwork for GPT-6 Astra

OpenAI

GPT-6 Astra

OpenAI's GPT-6 flagship, with top scores on computer use, browsing and software engineering.

Chat with GPT-6 Astra in Super AgentView pricing

What is GPT-6 Astra?

GPT-6 Astra is the flagship of OpenAI's GPT-6 series, announced on September 3, 2026.

OpenAI calls it the world's most intelligent and aligned model and reports state-of-the-art results on computer use, browsing, software engineering, cybersecurity, science and professional work, including 72.6% on OSWorld 2.0 in about 47% less time per task than GPT-5.6 Sol. It is also better at following templates and building well-structured slides, documents and spreadsheets. It is the first OpenAI model rated Critical for cybersecurity under OpenAI's Preparedness Framework, so it refuses advanced offensive security work and runs with misalignment monitoring.

Provider
OpenAI
Context window
1.05M tokens
Input
Text, Image
Reasoning
Low to max
Max output
128K tokens
Knowledge cutoff
Apr 30, 2026
Released
Sep 3, 2026

How GPT-6 Astra performs

Published by OpenAI

Results OpenAI reported for GPT-6 Astra. Each lab runs its own setup, so compare these with other models only inside the same table.

Benchmark results OpenAI published for GPT-6 Astra
BenchmarkScoreSetting
Agents' Last Exam59.3%best at any effort
OSWorld 2.072.6%offline set, partial score, best at any effort
Terminal-Bench 4.057.9%best at any effort
DeepSWE v1.174.1%best at any effort
FrontierCode 1.1 Main53.3%best at any effort
BrowseComp91.5%best at any effort
GPQA Diamond96.0%best at any effort
Humanity's Last Exam57.2%with tools, best at any effort
ARC-AGI-295.0%best at any effort
OpenAI MRCR v2, 8 needles96.3%512K to 1M tokens, best at any effort

Source: openai.com

What you can do with GPT-6 Astra

  • OpenAI's most capable GPT-6 model

  • 72.6% on OSWorld 2.0 at launch, the best computer-use score in OpenAI's comparison

  • Better at templates and structured slides, documents and spreadsheets

  • Uses substantially fewer output tokens than earlier models in several of OpenAI's evaluations

  • Room for 1.05M tokens of text and images

How to prompt GPT-6 Astra

  • Give it the end goal of the whole project, since Astra is built for long, multi-step work.
  • Attach your template or an example deck when the output has to match a format.
  • Use GPT-6 Luna for quick follow-ups and come back to Astra for the heavy steps.

Where GPT-6 Astra fits best

  • Research-heavy creative projects

  • Decks, reports and spreadsheets that follow a template

  • Complex plans with many moving parts

  • Hard analysis where quality comes first

How GPT-6 Astra compares

GPT-6 Astra compared with other chat models on Flashloop
ModelLMArena textLMArena codingContext windowReasoning
GPT-6 Astra1477#321543#91.05M tokensLow to max
Gemini 3.5 Flash1477#301508#621M tokensLow, medium, high
Claude Sonnet 5.51478#291528#261M tokensLow to max
GPT-6.1 Sol1480#271537#141.05M tokensLow to max
GLM-5.31481#261521#331M tokensLow, medium, high, max

Good to know before you start

  • Pricier per token than GPT-6.1 Sol, which OpenAI prices at a fifth of Astra's standard rates
  • No reasoning-off option for instant replies
  • Refuses advanced offensive security work, and misalignment monitoring can stop a task

Start chatting with GPT-6 Astra

Chat with GPT-6 Astra in Super Agent

  1. Select GPT-6 Astra

    Open Super Agent on Flashloop and choose GPT-6 Astra in the model picker.

  2. Enter your prompt

    Ask for what you need: an answer, a plan, or the images and videos you want the agent to make.

  3. Send and keep going

    Send the message. GPT-6 Astra answers in Super Agent and runs any generations you asked for.

Questions about GPT-6 Astra

Is GPT-6 Astra OpenAI's top model?

Yes. It is the flagship of the GPT-6 series, above GPT-6 Sol and GPT-6 Luna.

What is GPT-6 Astra best for?

OpenAI reports state-of-the-art results on computer use, browsing, software engineering, cybersecurity, science and professional work, and says it is better at producing structured slides, documents and spreadsheets.

What reasoning effort does GPT-6 Astra use in Super Agent?

It starts at medium, and you can choose low, medium, high, extra high or max.

What is GPT-6 Astra Pro?

The same model in OpenAI's pro reasoning mode, which does more work on each request. In ChatGPT, OpenAI offers it as GPT-6 Pro for difficult tasks and longer-running workflows.

More chat models on Flashloop

Browse all models