# GPT-6 Astra

OpenAI, released 2026-09-03

**Verdict: conditional.** Conditional: on Artificial Analysis's v5 Intelligence Index (27 October 2026) Astra (max) scores 57, one point behind Claude Opus 5.5 (max), at $10/$50 per 1M tokens short context; OpenAI's GPT-6 Astra safety overview, as we read it on 2026-09-04, reported decreased monitorability versus GPT-5.6 Sol. Last verified 2026-10-13.

## Pricing

- Input: $10 / 1M tokens
- Output: $50 / 1M tokens
- Context: 1.05M tokens; short-ctx standard rate

## Strengths

- Agentic coding and cyber defense. OpenAI reports Astra beats rival models on bug detection and code analysis tests
- Long-horizon end-to-end work: 1.05M-token context, 128K max output, reasoning effort selectable up to max
- Low-latency EU data-residency workloads. Fast mode is unavailable for Astra under EU data residency
- Cost-sensitive high-volume traffic. Standard short-context output is $50 per 1M tokens and Fast long-context reaches $150

## Sources

- [OpenAI: Responding to the next frontier of critical cyber capabilities](https://openai.com/index/responding-next-frontier-critical-cyber-capabilities/)
- [OpenAI API models: GPT-6 Astra specifications](https://developers.openai.com/api/docs/models)
- [OpenAI API changelog, 5 November 2026 (GPT-6 Cyber)](https://developers.openai.com/api/docs/changelog)
- [OpenAI API pricing (cyber models)](https://developers.openai.com/api/docs/pricing)
- [TechCrunch: OpenAI launches Astra, its powerful (and controversial) new model](https://techcrunch.com/2026/09/03/openai-launches-astra-its-powerful-and-controversial-new-model/)
- [OpenAI API docs: gpt-6-astra](https://developers.openai.com/api/docs/models/gpt-6-astra)

Canonical: https://neomanex.com/models/astra