GPT-6 Astra

Name: GPT-6 Astra model facts
Creator: OSMEXO

openai/gpt-6-astra

OpenAI
Family · GPT-6 Astra
Released 3 Sep 2026
File · Image · Text → Text
Reasoning · adjustable
Tools · JSON · Vision

Live on OSMEXO

Try now Request a discount

API setup & price →

Spec & list price

Context1M

Max output128K

Input / 1M$10.00

Output / 1M$50.005× input

Scores · standing

Intelligence56.9#3/489

LiveBench85.6#2/32

Coding77.1#4/196

Usage

Tokens / day39.4B▼ 12% 7d

Usage rank#55

Requests / day630.6K

Tool calls / req0.46

Per request · 30 days

Avg input45.8K3.1× median

Avg output5280.8× median

Cache hit90%1.3× median

Reasoning share43%0.8× median

Tool-call failures0.2%0.1× median

Consumption

Peak 459.8B 1 OctLow 29.8B 1 NovRank today #55

News & trends

404 mentions this week ▲ 677%this model · videos, posts, articles · 30 days

Trending · 14 days

Post-training 4B model with RL to fly Tesla Roadster 21% better than GPT-6 Astra12 points
Gemini 4 Carbon - The Worlds Smartest AI (Beats Opus 5.5 and GPT-6 Astra)70.0K views
DeepSeek, Moonshot used OpenAI's GPT-6 Astra to train latest models3 points
Kimi K3.1 против Opus 5,5 и GPT-6 Astra – Я ВСЕ ПРОВЕРИЛ88.5K views
GPT-6 Astra painted this on a real robot arm, then filmed and edited it, all from one chat874 upvotes
@emollick on X92.4K views
C-section animation: Claude Opus 5.5 vs GPT-6 Astra. Same prompt, different results.499 upvotes
@TheRundownAI on X7.6K views

News

Chinese labs now target OpenAI's GPT-6 Astra for distillationfork.zone · 1 Nov
Post-training 4B model with RL to fly Tesla Roadster 21% better than GPT-6 Astraskypilot.ai · 30 Oct
DeepSeek, Moonshot used OpenAI's GPT-6 Astra to train latest modelsreuters.com · 29 Oct
Retell uses OpenAI models to help power 100M monthly callsOpenAI · 27 Oct

Benchmarks

bar = share of the top score · Benchmark data: Artificial Analysis

GPQA Diamond#1 of 445

96.3%

Terminal-Bench 2.1#2 of 182

89.9%

Intelligence index#3 of 489

56.9

Coding index#4 of 196

77.1

Long-context reasoning#43 of 393

80.7%

Humanity's Last Exam#8 of 465

54.7%

Terminal-Bench 4.0#4 of 152

59.6%

SciCode#25 of 160

56.5%

Speed · median across 7 providers, latest day

Output speed#238/33448 tok/s1.4× slower than median

Fastest provider120 tok/sbest of 7 providers

First token#3023.42 s3.3× longer than median

First answer token#97164 sthinking included

#5 cheapest of 5 with a similar score

Value

Intelligence Index

Blended price per 1M tokens (log)

LiveBench suites

Browser91.3

Coding93.7

Data Analysis83.0

Design82.0

Long Agentic77.5

Reasoning92.7

Research85.2

▲ Strongest: Coding, Reasoning. ▼ Weakest: Long Agentic (78).

Technical details

Reasoning: on · medium
Tokenizer: GPT
Moderation: moderated

Reasoning effort

max
xhigh
high
medium · default
low

Request parameters 11

tools
tool_choice
response_format
structured_outputs
reasoning
reasoning_effort
include_reasoning
max_tokens
max_completion_tokens
seed

+1 more

verbosity

When to use

Best for

Science and expert-level reasoning
Coding agents on hard problems
Terminal-based automation
Very long document analysis
Image and file understanding

Think twice for

Design-heavy front-end generation
High-volume simple chat (cheaper options exist)
Self-hosting (closed weights)

Research papers

Same family