← Back
AWS ML Blog

OpenAI Codex on Amazon Bedrock

•14 min read•
#bedrock#llm#deployment
Level:Advanced
For:AI Engineers
✦TL;DR

OpenAI Codex can invoke Bedrock‑hosted OpenAI models such as GPT‑5.6 Sol from a developer workstation, and Amazon provides three rollout patterns—Native AWS Access, AgentCore Gateway, and LLM Gateway—to align inference with existing IAM, regional, and observability controls. The quick‑start repository supplies ready‑to‑run configurations for each pattern, including native IAM integration, an AgentCore‑managed gateway with Guardrails, and a custom LLM gateway (e.g., LiteLLM on ECS or Portkey). Selecting the appropriate path lets platform teams enforce least‑privilege access, telemetry via ADOT and CloudWatch, and model‑region approvals before scaling.

⚡ Key Takeaways

  • Model used in examples: **GPT‑5.6 Sol** (also mentions Sol, Terra, Luna availability per region).
  • Architecture option: **AgentCore Gateway** provides AWS‑managed routing and Bedrock Guardrails.
  • Trade‑off: **Native AWS Access** offers the lowest operational overhead, while custom LLM Gateway adds request‑path control at higher complexity.
  • Integration point: Use the **Native AWS Access quick start** from the open‑source “Guidance for Codex on AWS” repo to configure IAM Identity Center for the pilot.
  • Prerequisite: Verify model and region support via the **Supported models by AWS Region** list before deployment.
💡 Why It Matters

Engineers can securely expose Codex‑driven code assistance to internal teams without building a separate model‑access stack, leveraging existing AWS IAM and observability tooling to meet compliance and cost‑control requirements.

✅ Practical Steps

  1. Clone the “Guidance for Codex on AWS” repository and run the **Native AWS Access quick start** to provision IAM Identity Center permissions for Codex.
  2. Verify that **GPT‑5.6 Sol** (or Terra/Luna) is listed in the **Supported models by AWS Region** table for your target region.
  3. Deploy the chosen gateway (e.g., **AgentCore Gateway** or **LiteLLM on ECS**) and configure **ADOT** and **Amazon CloudWatch** for telemetry as described in the quick‑start scripts.

Want the full story? Read the original article.

Read on AWS ML Blog ↗

More like this

Introducing workload recovery and expanded infrastructure checks on Amazon SageMaker HyperPod

AWS ML Blog•#deployment

Introducing Unity Gateway Insights

Databricks Blog•#deployment

Introducing Gemini 4 Neon

Google DeepMind•#llm

DaisyRose-27B: Pushing the Frontier of Post-Training Efficiency

Hugging Face Blog•#llm

EXPLORE AI NEWS

Daily hand-picked stories on LLMs, RAG, agents and production AI — curated for engineers who ship.

BROWSE NEWS

GET THE WEEKLY DIGEST

Join engineers getting the Monday signal-over-noise AI breakdown. No spam, unsubscribe anytime.

LEARN AI ENGINEERING

Curated courses, research papers, repos and tutorials built for engineers leveling up in AI.

START LEARNING