Accelerating software development with OpenAI ChatGPT Codex on Amazon Bedrock

By: Sunil Bemarkar, Sr. Partner Solutions Architect – AWS
By: Garrett McCarthy, Partner Leader – OpenAI
By: Erik Anderson, Principal Solutions Architect – AWS

OpenAI

The pace of software development continues to accelerate, with organizations under increasing pressure to ship features faster, reduce technical debt, and do more with constrained engineering resources. According to Gartner, more than 80% of enterprises will have used generative AI APIs or deployed generative AI–enabled applications in production environments by 2026, up from less than 5% in 2023, signaling a fundamental shift in how software is built. Meanwhile, research from GitHub shows that developers using AI coding assistants complete tasks up to 55% faster while reporting significantly reduced cognitive load on repetitive work.

Yet for enterprises managing large codebases across distributed teams, merely adopting AI tools isn’t enough. Gartner predicts that by 2028, 75% of enterprise software engineers will use AI code assistants, up from less than 10% in early 2023. With the question of whether to adopt AI-assisted development resolved, the challenge becomes how to do so with security, governance, and operational controls that enterprise environments demand.

In this post, we explore how OpenAI ChatGPT Codex on Amazon Bedrock, OpenAI’s powerful coding agent for end-to-end software development, enables enterprises to adopt advanced AI-assisted development while maintaining the governance, security, and operational excellence that AWS provides. ChatGPT Codex is now available on Amazon Bedrock with OpenAI’s frontier models.

Scaling developer productivity without sacrificing quality

Enterprise software teams face a common set of challenges:

• Growing complexity – Modern applications span microservices, multi-cloud environments, and diverse technology stacks, making it harder for any single developer to comprehend the full system context.
• Talent constraints – The demand for experienced software engineers continues to outpace supply, and onboarding new developers to large codebases takes months.
• Quality and security requirements – Enterprises can’t trade speed for safety. Every line of code must meet compliance, security, and operational standards before reaching production.
• Technical debt accumulation – Teams often deprioritize refactoring and modernization under delivery pressure, creating a compounding burden that slows future development.

These challenges call for a new approach that augments human developers with AI agents capable of reasoning through complex problems at scale.

Introducing OpenAI ChatGPT Codex on Amazon Bedrock

ChatGPT Codex goes beyond straightforward code completion to work across the entire software development lifecycle (SDLC). Powered by OpenAI’s latest frontier models with advanced reasoning capabilities, Codex can:

• Understand full codebase context – Codex reasons across entire repositories, understanding dependencies, architectural patterns, and business logic to produce contextually appropriate solutions.
• Execute multistep coding tasks – From implementing new features to refactoring legacy code, Codex breaks down complex tasks into logical steps and executes them end to end.
• Generate tests and documentation – Codex can independently write unit tests, integration tests, and documentation aligned with your team’s conventions.
• Operate as a collaborative agent – Rather than replacing developers, Codex acts as a force multiplier that handles routine and complex coding tasks so engineers can focus on architecture, design, and innovation.

By making ChatGPT Codex available on Amazon Bedrock, enterprises gain access to OpenAI’s frontier reasoning models through fully managed, serverless API, thus eliminating the need to manage infrastructure while benefiting from the enterprise-grade capabilities of AWS services.

The case for AI-assisted development: What the research shows

The impact of AI coding assistants on developer productivity is well-documented by independent research:

• As mentioned in the first paragraph, a controlled study by GitHub research found that developers using AI coding assistants completed tasks 55% faster, averaging 1 hour 11 minutes compared to 2 hours 41 minutes without assistance.
• In an enterprise study with Accenture, GitHub found that AI-assisted developers saw an 84% increase in successful builds and a 15% rise in pull request merge rates, demonstrating improved code quality at scale.
• Productivity gains are becoming more measurable; Gartner insights reveal that 90% of software engineering leaders report visible improvements, logging a net average developer productivity gain of 19.3% from AI integration.

ChatGPT Codex builds on these foundations by going beyond code suggestions to deliver full agentic reasoning, understanding multifile context, executing complex multistep tasks, and collaborating with developers as an intelligent partner rather than an autocomplete tool.

Why Amazon Bedrock for OpenAI ChatGPT Codex?

Amazon Bedrock provides the foundation for enterprises to adopt generative AI responsibly, at scale. Running ChatGPT Codex on Amazon Bedrock delivers several critical advantages, including zero operator access, enterprise security and governance, operational visibility and monitoring, and unified billing.

The next-generation inference engine of Amazon Bedrock implements a zero operator access (ZOA) design that intentionally eliminates the technical means for AWS or OpenAI operators to access customer data, including inference prompts and completions.

AWS Identity and Access Management (IAM) integration provides fine-grained access control, limiting the ability to invoke Codex capabilities to only authorized users and systems. For customers who require private connectivity, AWS PrivateLink can be configured so traffic between your virtual private cloud (VPC) and Amazon Bedrock stays on the AWS network instead of traversing the public internet.

Amazon CloudWatch provides real-time metrics on usage, latency, and error rates, so your operations teams can maintain visibility into AI-assisted development workflows. AWS CloudTrail logs every API call for compliance auditing and security investigations.

You can enjoy consolidated billing through your existing AWS account alongside other AWS workloads. Pay-per-token pricing has no seat licenses or upfront commitments. Existing Amazon Bedrock customers can access OpenAI’s latest frontier models in one place.

Business outcomes: What enterprises can expect

Organizations adopting ChatGPT Codex on Amazon Bedrock can realize measurable impact across their software development lifecycle, as outlined in the following table.

Who benefits from OpenAI ChatGPT Codex on Amazon Bedrock?

ChatGPT Codex on Amazon Bedrock serves a broad spectrum of enterprise personas:

• Engineering leaders and CTOs seeking to scale development capacity without proportional headcount growth while maintaining security and compliance.
• Individual developers looking for an intelligent coding partner that understands their code base, reduces context-switching, and handles tedious tasks.
• Executive decision-makers evaluating generative AI investments who need a solution that integrates with existing AWS infrastructure, governance, and procurement workflows.
• Other technical teams (product managers, data analysts, technical writers) who work alongside engineering and want an agent that can read the code base, query documentation, and automate repetitive tasks without writing code.

Getting started

ChatGPT Codex runs locally on your machine using the command line interface (CLI), desktop app, or integrated development environment (IDE) extension and routes model requests to Amazon Bedrock using AWS authentication.

To begin using ChatGPT Codex on Amazon Bedrock, complete the following steps:

1. To download and install, visit Codex. You can download the Codex CLI, Codex App, or the Codex VS Code extension and get started with Amazon Bedrock for model inference. Confirm you’re on Codex CLI 0.128.0 or later, the Codex desktop app, or Visual Studio Code extension 26.429.30905 or later. Earlier builds don’t support Amazon Bedrock as a model provider.
2. Configure authentication. Codex supports two Amazon Bedrock authentication paths, the Amazon Bedrock API key or AWS SDK credential chain, which gives teams the flexibility to choose the approach that best fits their existing security posture and infrastructure setup. When you set a value for AWS_BEARER_TOKEN_BEDROCK, Codex uses that first. If that doesn’t exist, it falls back to the AWS SDK credential chain. You can set the bearer token in the environment as:

export AWS_BEARER_TOKEN_BEDROCK=<your-bedrock-api-key>

3. Set Amazon Bedrock as your provider, your preferred Region, and the model ID in the Codex configuration in ~/.codex/config.toml, which is required for Bedrock-API-key authentication as shown in the following code. Confirm that you provide models that match the model Region compatibility. As of this writing, openai.gpt-5.6 (Sol, Terra, and Luna), openai.gpt-5.5, and openai.gpt-5.4 are supported.

model = "openai.gpt-5.5"
model_provider = "amazon-bedrock"
model_reasoning_effort = "medium"
[model_providers.amazon-bedrock.aws]
region = "us-east-2"

4. Restart the desktop app or VS Code extension after changing ~/.codex/config.toml. For Codex CLI, type codex in the terminal to get started with Codex on Amazon Bedrock.

All model inference is routed through the Responses API on Amazon Bedrock, with pay-per-token pricing that requires no seat licenses or per-developer commitments, and data residency stays within the Amazon Bedrock Region you select.

To learn more about how to configure ChatGPT Codex to use OpenAI models on Amazon Bedrock, visit Use Codex with Amazon Bedrock.

Conclusion

The convergence of frontier AI reasoning and enterprise cloud infrastructure changes how organizations build software and perform knowledge work. With Gartner forecasting that 75% of enterprise engineers will use AI code assistants by 2028, the early adopters who invest in secure, governed AI development will be better positioned as adoption scales.ChatGPT Codex on Amazon Bedrock combines OpenAI’s advanced coding intelligence with AWS security, governance, and operational maturity. Enterprises can confidently adopt AI-assisted development that accelerates delivery, reduces technical debt, empowers developers with an AI collaborator, and still meet enterprise requirements.

To get started, explore OpenAI models on Amazon Bedrock and follow the guide to use ChatGPT Codex with Amazon Bedrock.

OpenAI – AWS Partner Spotlight

OpenAI is an AWS Advanced Technology Partner offering frontier AI models and platforms that help organizations solve complex problems. ChatGPT Codex, their agentic coding platform, brings advanced reasoning capabilities to software development workflows.

Contact OpenAI | Partner Overview