Model
GPT-6 Astra
GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.
Compare providers (sorted by listed input price)
21 providers list GPT-6 Astra. The cheapest full-context option is DigitalOcean at $10/$50 per M.
Best price history
Daily lowest published price across live providers. A lower line is better for the buyer.
- Best input /M
- $10unchanged
- Best output /M
- $50unchanged
- Provider coverage
- 25
- 15 at the start · 9 snapshots
Published best prices are unchanged.
9 daily snapshots collected from Sep 5, 2026 through Sep 14, 2026. A price chart will appear after the first real move.
Price badge
https://www.vaanalytics.in/badge/openai/gpt-6-astra.svgLive SVG, regenerated on every hourly sync — paste it into a README to always show the current cheapest listed price. Updates when prices change; no build step needed on your side.
Recent changes
Changelog →- new modelvia llmgateway-providers · Sep 13, 2026
- capabilityvia amazon-bedrock · Sep 10, 2026
- structured_output: false → true
- capabilityvia amazon-bedrock · Sep 10, 2026
- structured_output: false → true
- new modelvia amazon-bedrock · Sep 9, 2026
- new modelvia requesty · Sep 9, 2026
- new modelvia 302ai · Sep 9, 2026
In the news
All news →- OpenAI claims GPT-6 Astra is an ethereal 'Alien Mind' with AGI-like qualities
tomshardware.com· 5d ago - GPT-6 Astra Launch: $10/$50 Pricing, How to Try [2026]
shattered.io· 6d ago - GPT-6 Astra: A new generation of intelligence
openai.com· 4d ago
- Take on your most ambitious work with GPT-6 Astra ...
aws.amazon.com· 5d ago
Benchmarks
- Agents' Last Examreport ↗Unclassified source59.3
- OSWorld ↗(partial score)report ↗Unclassified source72.6
- ScreenSpot-Pro(accuracy)report ↗Unclassified source92.7
- AutomationBench(success rate)report ↗Unclassified source41.4
- BenchCAD(geometric overlap)report ↗Unclassified source95.9
- BrowseComp ↗(accuracy)report ↗Unclassified source91.5
- Terminal-Bench ↗(success rate)report ↗Unclassified source57.9
- DeepSWE(resolve rate)report ↗Unclassified source74.1
- FrontierCode(score)report ↗Unclassified source64.5
- Terminal-Bench Science ↗(success rate)report ↗Unclassified source64.6
- FrontierMath ↗(accuracy)report ↗Unclassified source97.6
- GPQA Diamond ↗(accuracy)report ↗Unclassified source96
- Humanity's Last Exam ↗(accuracy)report ↗Unclassified source57.2
- ExploitBench(success rate)report ↗Unclassified source100
- SRE-Bench(success rate)report ↗Unclassified source88
- ARC-AGI-3 ↗(RHAE)report ↗Unclassified source99.9
- Artificial Analysis Intelligence Index ↗(index score)report ↗Independent61
- Artificial Analysis Coding Agent Index ↗(index score)report ↗Independent67
Scores a lab published about its own model are claims, not measurements. Rankings elsewhere on this site use independently measured scores only.
Prices are per million tokens (USD). “—” means the provider does not publicly list a price for this model. Data from models.dev; verify with the provider before purchasing.