GPT-6 Astra API Pricing Calculator & Benchmarks

GPT-6 Astra: OpenAI’s most capable frontier model for advanced reasoning, coding, computer use, research, science, and professional workflows.

GPT-6 Astra is a multimodal model from OpenAI, released 3 September 2026. It pairs a 1,050,000-token context window with up to 128,000 tokens of output, taking text, image and file input and returning text. Training data ends 30 April 2026, and the model reasons step by step. It costs between $5.00 and $40.00 per million input tokens.

Last updated Oct 27, 2026

Never miss an AI pricing or launch update

Get notified about new model launches, pricing changes, provider updates, deprecations, and cost-saving insights.

πŸ”’ We respect your privacy. Unsubscribe anytime.

Model Details

Released
Sep 3, 2026
Knowledge Cutoff
Apr 30, 2026
Context Length
1,050,000
Max Output
128,000
Modalities
Text Image File → Text
Capabilities
Tool use Structured outputs Web search Prompt caching Code execution Computer use

Where is GPT-6 Astra a perfect fit?

GPT-6 Astra is OpenAI’s flagship frontier model, combining advanced reasoning with computer use, coding, browsing, scientific research, cybersecurity, and complex professional workflows across applications and autonomous agents.

- Complex end-to-end reasoning and knowledge work
- Agentic software engineering and repository-scale coding
- Computer-use automation across websites and desktop applications
- Scientific and mathematical research
- Professional document, spreadsheet, and presentation creation
- Browser-based research and multi-step workflows
- Cybersecurity research and analysis
- Multi-agent applications requiring sustained reasoning and tool use

Quick Model Estimate

(USD 10.0000 per 1M tokens)
(USD 50.0000 per 1M tokens)

Your GPT-6 Astra Cost Estimate

πŸ’° Total Cost

β€”

for 1000 input + 1000 output tokens

πŸ“₯ Input (1000 Γ— $10.000000) β€”
πŸ“€ Output (1000 Γ— $50.000000) β€”

Cost Breakdown

πŸ“₯ Input πŸ“€ Output

Prices are watched for changes and checked against the provider's own page.

Pricing

Provider ↕
Modality ↕
Service Tier ↕
Input Price
(per 1M tokens)
↕
Output Price
(per 1M tokens)
↕
Cached Input
(per 1M tokens)
↕
Cached Output
(per 1M tokens)
↕
Context Size ↕
View
OpenAI LogoOpenAITextStandard$10.0000$50.0000$1.0000$12.5000Up to 272,000 tokens→
OpenAI LogoOpenAITextStandard$20.0000$75.0000$2.0000$25.0000272,000+ tokens→
OpenAI LogoOpenAITextBatch$5.0000$25.0000$0.5000$6.2500Up to 272,000 tokens→
OpenAI LogoOpenAITextBatch$10.0000$37.5000$1.0000$12.5000272,000+ tokens→
OpenAI LogoOpenAITextFlex$5.0000$25.0000$0.5000$6.2500Up to 272,000 tokens→
OpenAI LogoOpenAITextFlex$10.0000$37.5000$1.0000$12.5000272,000+ tokens→
OpenAI LogoOpenAITextPriority$20.0000$100.0000$2.0000$25.0000Up to 272,000 tokens→
OpenAI LogoOpenAITextPriority$40.0000$150.0000$4.0000$50.0000272,000+ tokens→
Perplexity LogoPerplexityTextStandard$10.0000$50.0000$1.0000N/AUp to 272,000 tokens→
Perplexity LogoPerplexityTextStandard$20.0000$75.0000$2.0000N/A272,000+ tokens→

Benchmarks

Scores from standardized evaluations by Artificial Analysis and Design Arena. Higher is better β€” the indices summarize overall ability, while the detailed scores break down performance on individual benchmarks.

Artificial Analysis

Higher is better Β· benchmarked by Artificial Analysis
Index scores by variant
Detailed scores
52.7
Intelligence Index
Overall intelligence across reasoning, knowledge & math evals
Better than 90% of 32 models
76.9
Coding Index
Coding ability across software-engineering evals
Better than 80% of 16 models
51
Agentic Index
Tool use & multi-step agent task performance
Better than 75% of 13 models
Reasoning
GPQA Diamond 96.1%

Graduate-level questions in biology, chemistry & physics

Humanity's Last Exam 54.7%

Expert-level questions across many academic domains

CritPt 31.7%

Unpublished physics research reasoning problems

AA-LCR 80.7%

Long-context reasoning across large inputs

Coding
SciCode 56.5%

Research-level scientific coding tasks

Knowledge
GDPval 53.7%

Economically valuable, real-world knowledge work

Omniscience Accuracy 62.6%

Breadth of factual knowledge across domains

Omniscience Non-Hallucination 48.7%

How reliably the model avoids fabricated answers

All variants
Variant GPQA Diamond Humanity's Last Exam AA-LCR SciCode CritPt GDPval Omniscience Accuracy Omniscience Non-Hallucination
GPT-6 Astra (Max) 96.1% 54.7% 80.7% 56.5% 31.7% 53.7% 62.6% 48.7%
GPT-6 Astra (Xhigh) 96.3% 54.6% 80% 55.7% 31.4% 52.8% 61.9% 51.7%
GPT-6 Astra (High) 94.9% 53.1% 80% 55.4% 28.9% 51.2% 61.1% 55.2%
GPT-6 Astra (Medium) 93.9% 52.7% 79.7% 54.2% 29.1% 49.3% 60.6% 53.5%
GPT-6 Astra (Low) 93.1% 49.2% 80% 54.1% 26.3% 44.5% 59.5% 53.1%

FAQs about GPT-6 Astra

How much does GPT-6 Astra cost per 1M tokens?

GPT-6 Astra costs between $5.00 and $40.00 per million input tokens, and between $25.00 and $150.00 per million output tokens. The range covers Standard, Batch, Flex and Priority pricing.

What does a typical workload cost with GPT-6 Astra?

1,000 requests of 2,000 input and 500 output tokens each β€” 2,000,000 input and 500,000 output tokens in total β€” costs $22.50 with GPT-6 Astra at its lowest rates. Use the calculator on this page for your own volumes.

Which provider is cheapest for GPT-6 Astra?

OpenAI has the lowest input rate for GPT-6 Astra, at $5.00 per million tokens. Across all providers: OpenAI from $5.00 and Perplexity from $10.00 per million input tokens.

Is there cheaper GPT-6 Astra pricing than the standard rate?

Yes. Batch pricing for GPT-6 Astra starts at $5.00 per million input tokens against $10.00 on the standard tier, about 50% lower.

What is GPT-6 Astra's context window?

GPT-6 Astra has a context window of 1,050,000 tokens, and returns up to 128,000 tokens in a single response.

What input types does GPT-6 Astra support?

GPT-6 Astra accepts text, image and file input, and returns text.

What is GPT-6 Astra's knowledge cutoff?

GPT-6 Astra's training data runs to 30 April 2026. It has no built-in knowledge of events after that date.