--- title: "OpenAI Brings GPT-5.5 and Codex to AWS, Breaking Microsoft's Grip" subtitle: "GPT-5.5, GPT-5.4, and Codex hit Amazon Bedrock GA, giving AWS-native enterprises a direct path to OpenAI's best models without leaving their cloud" url: "https://alphasignal.ai/news/openai-brings-gpt-5-5-and-codex-to-aws-breaking-microsoft-s-grip" author: "AlphaSignal Newsroom" author_url: "https://alphasignal.ai/editorial-team" company: "OpenAI" company_url: "https://alphasignal.ai/company/openai" published: "2026-06-01T21:44:13.000Z" updated: "2026-06-16T20:36:03.846Z" type: "News" topics: - "Llms" - "Api" subtopics: - "Vision Language" upvotes: 5424 read_time: "6 min" source_url: "https://openai.com/index/openai-frontier-models-and-codex-are-now-available-on-aws/" comments: 0 --- # OpenAI Brings GPT-5.5 and Codex to AWS, Breaking Microsoft's Grip > GPT-5.5, GPT-5.4, and Codex hit Amazon Bedrock GA, giving AWS-native enterprises a direct path to OpenAI's best models without leaving their cloud For years, if you wanted OpenAI's best models in a production enterprise environment, you had one real option: go through Azure. That just changed. [OpenAI's frontier models and Codex are now generally available on Amazon Bedrock](https://openai.com/index/openai-frontier-models-and-codex-are-now-available-on-aws/), opening a second major cloud distribution channel for OpenAI capabilities and putting real pressure on Microsoft's historically exclusive grip on enterprise OpenAI access. ## What's actually shipping One month after the expanded partnership announcement, GPT-5.5, GPT-5.4, and Codex are now generally available on Amazon Bedrock, giving customers access to frontier models and the OpenAI coding agent for software development. There are two distinct products here: - **OpenAI models on Amazon Bedrock** -- GPT-5.5 and GPT-5.4 available through the Responses API, callable with the same AWS SDK and IAM controls you already use. Both run on Bedrock's next-generation inference engine, built for high performance, reliability, and security, with pricing that matches OpenAI rates and no additional fees. - **Codex on Amazon Bedrock** -- available through the Codex App, CLI, and IDE integrations, with all inference routed through Amazon Bedrock and the same security protections -- including IAM, VPC isolation, and encryption -- that customers already use on AWS. Codex will be available through Bedrock via the Codex CLI, desktop app, and VS Code extension. JetBrains and Xcode are also supported. Usage of both OpenAI models and Codex can be applied toward existing AWS cloud commitments, which is a significant procurement win for teams that already have AWS spend commitments to burn through. ## The deal behind the deal This GA launch is the first visible product milestone of a much larger strategic bet. OpenAI and Amazon announced a multi-year strategic partnership, with Amazon committing to invest $50 billion in OpenAI -- starting with an initial $15 billion investment, followed by another $35 billion when certain conditions are met. The compute side of the deal is equally significant. OpenAI and AWS are expanding their existing $38 billion multi-year agreement by $100 billion over 8 years. OpenAI will consume 2 gigawatts of Trainium capacity through AWS infrastructure to support demand for Stateful Runtime Environment, Frontier, and other advanced workloads. That's a massive bet on Amazon's custom AI silicon -- Trainium -- which has historically lagged Nvidia GPUs and Google's TPUs. Critically, AWS will be the exclusive third-party cloud distribution provider for OpenAI Frontier, which enables organizations to build, deploy, and manage teams of AI agents. That exclusivity clause is the structural moat that makes this more than a simple model listing. ## Why procurement friction was the real blocker The rollout matters because procurement friction -- not raw model capability -- is what keeps many Fortune 500 teams from standardizing frontier stacks. Getting a new vendor approved for security review, compliance sign-off, and billing integration can take months inside large organizations. By routing OpenAI through Bedrock, teams skip that entire process. For enterprises, this removes one of the biggest barriers to AI adoption: getting frontier AI into production through existing security, compliance, procurement, billing, and governance workflows. Every call through Bedrock inherits IAM permissions, VPC and PrivateLink isolation, KMS encryption, and AWS CloudTrail audit logging automatically. Crucially, prompts and responses are not used to train models and are not shared with OpenAI. Early adopters include Amgen and Autodesk. GPT-5.5 and frontier models offer compelling advances in capability, quality, and consistency that matter in a field where the questions are complex and the standards for scientific accuracy and decision quality are exceptionally high. ## Who wins, who feels the squeeze The clearest winner is the large class of AWS-native enterprises that wanted OpenAI models but had no appetite to onboard Azure. They now get GPT-5.5 without a new cloud relationship, new billing setup, or new security review. Pricing on Bedrock matches OpenAI's first-party rates with no additional fees, and usage counts toward existing AWS cloud commitments. The pressure lands squarely on Microsoft. Azure OpenAI Service built its enterprise moat on being the only cloud where you could run OpenAI models under enterprise-grade controls. The partnership hit a legal roadblock almost immediately -- Microsoft's exclusive licensing agreement had previously prevented OpenAI from distributing its API-accessible products through competing cloud providers. That exclusivity is now clearly gone. Azure OpenAI remains the right answer for enterprises already deep in the Microsoft ecosystem that need the strongest compliance posture, but it no longer has a monopoly on the enterprise OpenAI story. Anthropic, which has its own deep AWS relationship and runs Claude models on Bedrock, faces a more nuanced situation. Bedrock now hosts both Claude and GPT-5.5 side by side, which is good for AWS but increases direct model competition within the same platform. ## What's technically new for builders A few details worth knowing if you're planning to integrate: - **Isolated inference queues** -- Bedrock's engine gives each customer their own queue with automated capacity management, so your latency stays predictable under load. - **Durable state** -- if hardware fails mid-request, the call resumes rather than restarting from scratch. This matters a lot for long-running agentic tasks. - **Pay-per-token, no seat licenses** -- Codex on Bedrock drops the per-developer seat model entirely. You pay for what you use, which makes it easier to roll out broadly without budget forecasting headaches. - **GovCloud support** -- teams can use Codex to write, review, debug, and modernize code within AWS infrastructure, including Commercial and GovCloud regions. - **Data residency** -- inference stays within your selected AWS region, which matters for regulated industries. ## What's coming next The GA launch is explicitly framed as the start of a broader rollout. Amazon Bedrock Managed Agents, powered by OpenAI, will make it fast to deploy production-ready OpenAI-powered agents on AWS. At the core are the latest OpenAI frontier models and the OpenAI agent harness, engineered for faster execution, sharper reasoning, and reliable steering of long-running tasks. The more interesting upcoming capability is Daybreak. The next phase will include future availability for Daybreak, OpenAI's cyber-focused work covering secure code review, threat modeling, patch validation, dependency risk analysis, detection, and remediation guidance. Routing cybersecurity tooling through AWS's existing governance rails is a natural fit for the security teams that would actually use it -- they're already operating inside those controls every day. The broader implication is structural: OpenAI is now a multi-cloud model provider, not just an Azure-adjacent product. Stateful developer environments are the next generation of how frontier models will be used, seamlessly enabling models to access elements like compute, memory, and identity. The joint Stateful Runtime Environment that OpenAI and AWS are co-developing will be the infrastructure layer that makes long-running, memory-aware agents practical at enterprise scale. That's the real product being built here -- the [GA launch of GPT-5.5 and Codex on Bedrock](https://aws.amazon.com/blogs/machine-learning/openai-models-and-codex-on-amazon-bedrock-are-now-generally-available/) is just the first chapter. ## Sources - [X post by @openai](https://x.com/openai/status/2061564502160892138) - [OpenAI frontier models and Codex are now available on AWS | OpenAI](https://openai.com/index/openai-frontier-models-and-codex-are-now-available-on-aws/) - [OpenAI models and Codex on Amazon Bedrock are now generally available | AWS](https://aws.amazon.com/blogs/machine-learning/openai-models-and-codex-on-amazon-bedrock-are-now-generally-available/) - [OpenAI and Amazon Announce Strategic Partnership | OpenAI](https://openai.com/index/amazon-partnership/) - [OpenAI and Amazon Announce Strategic Partnership | About Amazon](https://press.aboutamazon.com/2026/2/openai-and-amazon-announce-strategic-partnership) - [OpenAI Models and Codex Are Now Generally Available on AWS | Memeburn](https://memeburn.com/openai-models-and-codex-are-now-generally-available-on-aws/)