Industry story
GPT-5.4 and NVIDIA Nemotron 120B added to Kiro in AWS GovCloud
ai-in-adtech cloud-costs engineering
GovCloud AI access just got serious. AWS has added GPT-5.4 and NVIDIA Nemotron 120B to its Kiro coding assistant in GovCloud (US-West) — the restricted region where federal agencies actually run sensitive workloads. GPT-5.4 brings a 272K context window and full agentic workflow support via Bedrock; Nemotron activates only 12B of its 120B parameters at inference time, priced at a 0.25x credit multiplier. The real story is speed of access: cleared environments have historically lagged commercial cloud by years, and that gap is closing fast.
Full analysis
Amazon Web Services has added two new AI models to its Kiro IDE and CLI coding assistant within the AWS GovCloud (US-West) Region, which is a restricted cloud environment for U.S. government workloads. OpenAI GPT-5.4 is available for complex reasoning, coding, and multi-step agentic workflows (tasks where an AI model autonomously plans and executes a sequence of actions), running on Amazon Bedrock's inference engine with a 272K token context window and isolated, durable execution queues. NVIDIA Nemotron 3 Super 120B — a hybrid mixture-of-experts model (an architecture that activates only a subset of its parameters at inference time for efficiency) — activates just 12B of its 120B parameters, offering fast and cost-efficient inference on agentic tasks with a 256K context window and a low 0.25x credit multiplier.
Comments