Get began with OpenAI GPT-5.5, GPT-5.4 fashions, and Codex on Amazon Bedrock

As we previewed in What’s Subsequent with AWS 2026, we’re asserting the final availability of OpenAI GPT-5.5, GPT-5.4 fashions, and Codex on Amazon Bedrock, supplying you with entry to frontier fashions and a coding agent for software program improvement.

Based on OpenAI, GPT-5.5 and GPT-5.4 fashions are glorious for coding, reasoning, agentic workflows, and sophisticated skilled work. You need to use GPT-5.5 for the toughest buyer workloads and GPT-5.4 for the most effective price-performance. You may name them via Responses API on Amazon Bedrock’s next-generation inference engine constructed for top efficiency, reliability, and safety.

Codex is the OpenAI coding agent for AI-powered software program improvement. Based on OpenAI, greater than 4 million builders use Codex each week to jot down, refactor, debug, take a look at, and validate code throughout giant codebases. With GPT-5.5 powering inference, Codex introduces a brand new class of intelligence optimized for complicated, long-horizon developer workflows. You need to use the Codex App, the Codex CLI, and IDE integrations with Visible Studio Code, JetBrains, and Xcode, with all mannequin inference routed via the Responses API on Amazon Bedrock.

For patrons with information residency necessities, all processing stays throughout the Bedrock Area you choose. You pay per token with no seat licenses and no per-developer commitments.

GPT-5.5 and GPT-5.4 fashions on Bedrock in motion

You may entry the mannequin programmatically utilizing the OpenAI Responses API to name the bedrock-mantle endpoints via the OpenAI SDK, command-line instruments equivalent to curl.

Let’s begin with OpenAI SDK for Python. Set up OpenAI SDK.

pip set up -U openai

Set the surroundings variables for authentication.

export OPENAI_BASE_URL="https://bedrock-mantle.us-east-2.api.aws/openai/v1"
export OPENAI_API_KEY=""
export BEDROCK_OPENAI_MODEL_ID="openai.gpt-5.5"

Here’s a pattern Python code to name GPT-5.5 mannequin on Bedrock:

import os
from openai import OpenAI
 
shopper = OpenAI(
    base_url=os.environ["OPENAI_BASE_URL"],
    api_key=os.environ["OPENAI_API_KEY"],
)
 
response = shopper.responses.create(
    mannequin=os.environ["BEDROCK_OPENAI_MODEL_ID"],
    enter=[
        {
            "role": "developer",
            "content": "You are a software engineer with excellent AWS cloud knowledge. Be concise and practical.",
        },
        {
            "role": "user",
            "content": "Design a distributed architecture on AWS in Python that should support 100k requests per second across multiple geographic regions.",
        },
    ],
    reasoning={"effort": "medium"},
    textual content={"verbosity": "low"},
)
 
print(response.output_text)

You may name straight the mannequin endpoint utilizing curl.

curl "$OPENAI_BASE_URL/responses" 
  -H "Content material-Kind: software/json" 
  -H "Authorization: Bearer $OPENAI_API_KEY" 
  -d '{
    "mannequin": "openai.gpt-5.5",
    "enter": [
      {
        "role": "developer",
        "content": "You are a software engineer with excellent AWS cloud knowledge."
      },
      {
        "role": "user",
        "content": "Design a distributed architecture on AWS in Python that should support 100k requests per second across multiple geographic regions."
      }
    ],
    "reasoning": {"effort": "medium"},
    "textual content": {"verbosity": "low"}
  }'

You need to use the Responses API if you wish to use model-managed multi-turn state, want hosted instruments, perform instruments, or richer device orchestration, and run background or long-running work. To study extra, go to the OpenAI Cookbook Responses examples and getting began information.

Utilizing OpenAI Codex with GPT-5.5 on Amazon Bedrock

You may obtain Codex CLI, Codex App or Codex VS Code extension and get began with the Bedrock for mannequin inference. Codex helps two Bedrock authentication pathways: Amazon Bedrock API key or AWS SDK credential chain. In the event you set AWS_BEARER_TOKEN_BEDROCK, Codex makes use of it first; in any other case Codex falls again to AWS SDK credential chain.

Set AWS_BEARER_TOKEN_BEDROCK within the surroundings that Codex will learn:

export AWS_BEARER_TOKEN_BEDROCK=

Then, configure your most well-liked Area and set the mannequin ID to openai.gpt-5.5 in ~/.codex/config.toml, which is required for Bedrock API-key authentication. You may as well select openai.gpt-5.4, openai.gpt-oss-120b, or openai.gpt-oss-20b. For the desktop app or VS Code extension, put any surroundings variables the app wants in ~/.codex/.env.

mannequin = "openai.gpt-5.5"
model_provider = "amazon-bedrock"
[model_providers.amazon-bedrock.aws]
area = "us-east-2"

Restart the desktop app or VS Code extension after altering ~/.codex/config.toml or ~/.codex/.env. In Codex CLI, it’s best to see a /standing tab that appears like this:

In Codex App, you should utilize GPT-5.5 mannequin via Amazon Bedrock inference.

To study extra about the right way to configure Codex to make use of OpenAI fashions on Amazon Bedrock, go to Use Codex with Amazon Bedrock.

Issues to know

Let me share some vital technical particulars that I feel you’ll discover helpful.

• Mannequin latency: OpenAI mannequin data positions GPT-5.5 as quick and GPT-5.4 as medium pace, however customer-perceived latency is determined by reasoning effort, output size, device calls, background mode, Area, quotas, throttling, immediate dimension, and cache hits. Begin GPT-5.5 at medium effort. Begin GPT-5.4 with effort set explicitly somewhat than counting on its none default.
• Scaling and capability: Bedrock’s new inference engine is designed to quickly provision and serve capability throughout many various fashions. When accepting requests, we prioritize preserving regular state workloads working, and ramp utilization and capability quickly in response to adjustments in demand. During times of excessive demand, requests are queued, somewhat than rejected.

Now out there

OpenAI GPT fashions and Codex on Amazon Bedrock can be found in the present day: GPT-5.5 mannequin within the US East (Ohio) Area, GPT-5.4 mannequin within the US East (Ohio) and US West (Oregon) Areas. Examine the full record of Areas for future updates. To study extra, go to the OpenAI on Amazon Bedrock web page and the Amazon Bedrock pricing web page.

Give GPT-5.5, GPT-5.4 fashions, and Codex on Amazon Bedrock a attempt in the present day and ship suggestions to AWS re:Publish for Amazon Bedrock or via your ordinary AWS Assist contacts.

— Channy

Up to date on June, 1, 2026 – The GPT fashions now assist the Responses API solely on Amazon Bedrock, and console assist is coming quickly.