AWS Machine Learning Blog•September 23, 2026•Published September 23, 2026

Use open weight models as your AI coding agent with Amazon Bedrock

OpenCode lets you run a private, cost‑effective AI coding agent on Amazon Bedrock’s open‑weight models, keeping data in‑house and avoiding per‑seat fees.

Why Now

Amazon Bedrock now supports open‑weight models, enabling developers to run AI coding agents privately and flexibly, a shift driven by cost and data‑residency concerns.

What Happened

OpenCode is an open‑source, terminal‑native coding agent written in Go that can connect to over 75 LLM providers, including Bedrock. When paired with Bedrock’s open‑weight models, it runs locally while inference stays within the user’s AWS account, eliminating infrastructure management and per‑seat charges. Bedrock offers three pricing tiers—Priority, Standard, and Flex—allowing users to match cost to workload and keep data residency compliant with programs like HIPAA and GDPR.

Why It Matters

This setup lets enterprises keep proprietary code and prompts inside their AWS environment, reducing compliance risk and potentially cutting multi‑million‑conversation costs by switching to cheaper open‑weight models. It also gives teams the flexibility to swap models per task, improving performance and cost efficiency for large‑scale agentic workflows.

The Limitation

The article focuses on Bedrock’s open‑weight offerings and OpenCode; real‑world performance and cost savings depend on specific workloads and may vary from the reported examples.

What You Can Do

Try OpenCode with Bedrock’s open‑weight models on a small project to benchmark cost and latency against your current coding assistant.

Source

Read original source

Why we picked this

Core AI content: introduces a new AI coding agent leveraging open-weight models on Amazon Bedrock, detailing configuration and usage.