- Introduces UltraFast mode for GPT-6 Astra on Amazon Bedrock, delivering up to 6× faster inference (~300 tokens/sec).
- Provides a premium speed tier for latency‑sensitive workloads such as real‑time coding assistants, interactive agents, and customer‑facing experiences.
- Available via the Bedrock console or APIs with built‑in AWS security, governance, and audit controls.