GPT‑6 Astra is being gradually released to a limited set of organizations, with broader availability expected soon through ChatGPT Plus, Pro, Business, Enterprise, the OpenAI API, and AWS.
The model is priced at $10 per million input tokens and $50 per million output tokens, aligning with Claude Fable 5.1. It appears to outperform Fable on most of OpenAI's self-reported benchmarks, notably achieving 99.9% on the ARC-AGI 3 benchmark.
Astra's performance on security tasks is notable, scoring 100% on ExploitBench and 99.2% on reverse engineering tasks, indicating strong security capabilities. It also excels in long context processing, handling up to 1 million tokens with high accuracy.
While Astra scores well on the Intelligence Index, it remains behind Claude Fable 5.1 and Meta’s Muse Spark 1.3. It leads in coding efficiency, offering better cost-performance than Claude Fable 5, making it a competitive option for tasks requiring long contexts and security.