亚马逊云科技宣布在其Amazon Bedrock和Amazon SageMaker JumpStart平台上线两款OpenAI开源权重模型gpt-oss-120b和gpt-oss-20b。这两款模型在代码编写、科学分析及数学推理方面表现出色，支持128K上下文窗口和可调的推理级别，并可集成外部工具以增强功能，适用于构建需要高度控制基础设施和数据的AI应用。用户可通过Bedrock的OpenAI兼容端点或SageMaker JumpStart快速评估、部署和定制这些模型，从而加速AI创新。

🌟 AWS在Amazon Bedrock和SageMaker JumpStart平台引入了OpenAI的gpt-oss-120b和gpt-oss-20b两款开源权重模型，旨在为开发者和企业提供更多先进的AI模型选择，以推动业务发展。

💡 这两款模型专为文本生成和推理任务设计，在代码编写、科学分析和数学推理方面性能优越，可与领先模型媲美。它们支持128K的超长上下文窗口，并提供可调节的推理级别（低/中/高）以满足不同用例需求。

🛠️ 用户可以通过Amazon Bedrock的OpenAI兼容端点或SageMaker JumpStart轻松访问和使用这些模型。Bedrock提供无缝的模型实验、混搭和切换能力，而SageMaker JumpStart则支持快速评估、定制和生产部署。

🚀 模型支持外部工具集成，可用于构建代理工作流，例如使用Strands Agents框架。这为开发更复杂的AI应用提供了灵活性和强大的功能扩展性，用户可以根据自身需求进行修改、适配和定制。

<section class="blog-post-content lb-rtxt"><table id="amazon-polly-audio-table"><tbody><tr><td id="amazon-polly-audio-tab"><p></p></td></tr></tbody></table><p>AWS is committed to bringing you the most advanced <a href="https://aws.amazon.com/what-is/foundation-models/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;foundation models (FMs)</a> in the industry, continuously expanding our selection to include groundbreaking models from leading AI innovators so that you always have access to the latest advancements to drive your business forward.</p><p>Today, I am happy to announce the availability of two new <a href="https://aws.amazon.com/bedrock/openai/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;OpenAI models with open weights</a> in <a href="https://aws.amazon.com/bedrock/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;Amazon Bedrock</a> and <a href="https://aws.amazon.com/sagemaker-ai/jumpstart/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;Amazon SageMaker JumpStart</a>. OpenAI <a href="https://openai.com/index/introducing-gpt-oss&quot;&gt;&lt;strong&gt;gpt-oss-120b&lt;/strong&gt; and <strong>gpt-oss-20b</strong></a> models are designed for text generation and reasoning tasks, offering developers and organizations new options to build AI applications with complete control over their infrastructure and data.</p><p>These open weight models excel at coding, scientific analysis, and mathematical reasoning, with performance comparable to leading alternatives. Both models support a 128K context window and provide adjustable reasoning levels (low/medium/high) to match your specific use case requirements. The models support external tools to enhance their capabilities and can be used in an agentic workflow, for example, using a framework like <a href="https://strandsagents.com/&quot;&gt;Strands Agents</a>.</p><p>With Amazon Bedrock and Amazon SageMaker JumpStart, AWS gives you the freedom to innovate with access to hundreds of FMs from leading AI companies, including OpenAI open weight models. With our comprehensive selection of models, you can match your AI workloads to the perfect model every time.</p><p>Through Amazon Bedrock, you can seamlessly experiment with different models, mix and match capabilities, and switch between providers without rewriting code—turning <a href="https://aws.amazon.com/bedrock/model-choice/&quot;&gt;model choice</a> into a strategic advantage that helps you continuously evolve your AI strategy as new innovations emerge. At launch, these new models are available in Bedrock via an OpenAI compatible endpoint. You can point the OpenAI SDK to this endpoint or use the Bedrock <a href="https://docs.aws.amazon.com/bedrock/latest/APIReference/API_runtime_InvokeModel.html&quot;&gt;InvokeModel&lt;/a&gt; and <a href="https://docs.aws.amazon.com/bedrock/latest/APIReference/API_runtime_Converse.html&quot;&gt;Converse API</a>.</p><p>With SageMaker JumpStart, you can quickly evaluate, compare, and customize models for your use case. You can then deploy the original or the customized model in production with the SageMaker AI console or using the <a href="https://docs.aws.amazon.com/sagemaker/latest/dg/jumpstart-foundation-models-use-python-sdk.html&quot;&gt;SageMaker Python SDK</a>.</p><p>Let’s see how these work in practice.</p><p><strong>Getting started with OpenAI open weight models in Amazon Bedrock<br /></strong> In the <a href="https://console.aws.amazon.com/bedrock/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;Amazon Bedrock console</a>, I choose <strong>Model access</strong> from the <strong>Configure and learn</strong> section of the navigation pane. Then, I navigate to the two listed OpenAI models on this page and request access.</p><p><img class="aligncenter size-full wp-image-98753" src="https://d2908q01vomqb2.cloudfront.net/da4b9237bacccdf19c0760cab7aec4a8359010b0/2025/08/05/openai-gpt-model-access.png&quot; alt="Console screenshot" width="1064" height="366" /></p><p>Now that I have access, I use the <strong>Chat/Test</strong> playground to test and evaluate the models. I select <strong>OpenAI</strong> as the category and then the <strong>gpt-oss-120b</strong> model.</p><p><img class="aligncenter size-full wp-image-98754 c6" src="https://d2908q01vomqb2.cloudfront.net/da4b9237bacccdf19c0760cab7aec4a8359010b0/2025/08/05/openai-gpt-model-selection.png&quot; alt="Console screenshot" width="845" height="707" /></p><p>Using this model, I run the following sample prompt:</p><p><em>A family has $5,000 to save for their vacation next year. They can place the money in a savings account earning 2% interest annually or in a certificate of deposit earning 4% interest annually but with no access to the funds until the vacation. If they need $1,000 for emergency expenses during the year, how should they divide their money between the two options to maximize their vacation fund?</em></p><p>This prompt generates an output that includes the chain of thought used to produce the result.</p><p><img class="aligncenter wp-image-98776 size-full" src="https://d2908q01vomqb2.cloudfront.net/da4b9237bacccdf19c0760cab7aec4a8359010b0/2025/08/06/2025-openai-models-bedrock-chat-playground-1.jpg&quot; alt="" width="1064" height="536" /></p><p>I can use these models with the OpenAI SDK by configuring the API endpoint (base URL) and using an <a href="https://docs.aws.amazon.com/bedrock/latest/userguide/api-keys.html?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;Amazon Bedrock API key</a> for authentication. For example, I set this environment variables to use the US West (Oregon) AWS Region endpoint (<code>us-west-2</code>) and my Amazon Bedrock API key:</p><pre class="lang-bash">export OPENAI_API_KEY="&lt;my-bedrock-api-key&gt;"export OPENAI_BASE_URL="https://bedrock-runtime.us-west-2.amazonaws.com/openai/v1&quot;&lt;/pre&gt;&lt;p&gt;Now I invoke the model using the OpenAI Python SDK.</p><pre class="lang-python">client = OpenAI()response = client.chat.completion.create( messages=[{ "role": "user", "content": "Hello, how are you?" }], model="openai.gpt-oss-120b-1:0", stream=True)for item in response: print(item)</pre><p>To build an AI agent, I can choose any framework that supports the Amazon Bedrock API or the OpenAI API. For example, here’s the starting code for Strands Agents using the Amazon Bedrock API:</p><pre class="lang-python">from strands import Agentfrom strands.models import BedrockModelfrom strands_tools import calculatormodel = BedrockModel( model_id="openai.gpt-oss-120b-1:0")agent = Agent( model=model, tools=[calculator])agent("Tell me the square root of 42 ^ 3")</pre><p>I save the code (<code>app.py</code> file), install the dependencies, and run the agent locally:</p><pre class="lang-bash">pip install strands-agents strands-agents-toolspython app.py</pre><p>When I am satisfied with the agent, I can deploy in production using the <a href="https://aws.amazon.com/blogs/aws/introducing-amazon-bedrock-agentcore-securely-deploy-and-operate-ai-agents-at-any-scale/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;capabilities offered by Amazon Bedrock AgentCore</a>, including a fully managed serverless runtime and memory and identity management.</p><p><strong>Getting started with OpenAI open weight models in Amazon SageMaker JumpStart<br /></strong> In the <a href="https://console.aws.amazon.com/sagemaker/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;Amazon SageMaker AI console</a>, you can use OpenAI open weight models in the <a href="https://aws.amazon.com/sagemaker-ai/studio/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;SageMaker Studio</a>. The first time I do this, I need to set up a SageMaker domain. There are options to set it up for a single user (simpler) or an organization. For these tests, I use a single user setup.</p><p>In the <strong>SageMaker JumpStart</strong> model view, I have access to a detailed description of the <strong>gpt-oss-120b</strong> or <strong>gpt-oss-20b</strong> model.</p><p><img class="aligncenter size-full wp-image-98774" src="https://d2908q01vomqb2.cloudfront.net/da4b9237bacccdf19c0760cab7aec4a8359010b0/2025/08/06/2025-openai-models-sagemaker-js-model.jpg&quot; alt="" width="1248" height="560" /></p><p>I choose the <strong>gpt-oss-20b model</strong> and then deploy the model. In the next steps, I select the instance type and the initial instance count. After a few minutes, the deployment creates an endpoint that I can then invoke in SageMaker Studio and using any <a href="https://aws.amazon.com/tools/&quot;&gt;AWS SDKs</a>.</p><p><img class="aligncenter size-full wp-image-98771" src="https://d2908q01vomqb2.cloudfront.net/da4b9237bacccdf19c0760cab7aec4a8359010b0/2025/08/06/2025-openai-models-sagemaker-js.jpg&quot; alt="" width="1440" height="799" /></p><p>To learn more, visit <a href="https://aws.amazon.com/blogs/machine-learning/gpt-oss-models-from-openai-are-now-available-on-sagemaker-jumpstart/&quot;&gt;GPT OSS models from OpenAI are now available on SageMaker JumpStart</a> in the AWS Artificial Intelligence Blog.</p><p><strong>Things to know<br /></strong> The new <a href="https://aws.amazon.com/bedrock/openai/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;OpenAI open weight models</a> are now available in <a href="https://aws.amazon.com/bedrock/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;Amazon Bedrock</a> in the US West (Oregon) <a href="https://aws.amazon.com/about-aws/global-infrastructure/regions_az/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;AWS Region</a>, while <a href="https://aws.amazon.com/sagemaker-ai/jumpstart/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;Amazon SageMaker JumpStart</a> supports these models in US East (Ohio, N. Virginia) and Asia Pacific (Mumbai, Tokyo).</p><p>Each model comes equipped with full chain-of-thought output capabilities, providing you with detailed visibility into the model’s reasoning process. This transparency is particularly valuable for applications requiring high levels of interpretability and validation. These models give you the freedom to modify, adapt, and customize them to your specific needs. This flexibility allows you to fine-tune the models for your unique use cases, integrate them into your existing workflows, and even build upon them to create new, specialized models tailored to your industry or application.</p><p>Security and safety are built into the core of these models, with comprehensive evaluation processes and safety measures in place. The models maintain compatibility with the standard GPT-4 tokenizer.</p><p>Both models can be used in your preferred environment, whether that’s through the serverless experience of Amazon Bedrock or the extensive <a href="https://aws.amazon.com/ai/machine-learning/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;machine learning (ML)</a> development capabilities of SageMaker JumpStart. For information about the costs associated with using these models and services, visit the <a href="https://aws.amazon.com/bedrock/pricing/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;Amazon Bedrock pricing</a> and <a href="https://aws.amazon.com/sagemaker-ai/pricing/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;Amazon SageMaker AI pricing</a> pages.</p><p>To learn more, see the <a href="https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-openai.html&quot;&gt;parameters for the models</a> and the <a href="https://docs.aws.amazon.com/bedrock/latest/userguide/inference-chat-completions.html&quot;&gt;chat completions API</a> in the Amazon Bedrock documentation.</p><p>Get started today with OpenAI open weight models on AWS in the <a href="https://console.aws.amazon.com/bedrock/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;Amazon Bedrock console</a> or in <a href="https://console.aws.amazon.com/sagemaker/?trk=e61dee65-4ce8-4738-84db-75305c9cd4fe&amp;amp;sc_channel=el&quot;&gt;Amazon SageMaker AI console</a>.</p><p>– <a href="https://x.com/danilop&quot;&gt;Danilo&lt;/a&gt;&lt;/p&gt;&lt;/section&gt;&lt;aside id="Comments" class="blog-comments"><div data-lb-comp="aws-blog:cosmic-comments" data-env="prod" data-content-id="74aacf85-d20a-4c73-86ab-f41138fba5ca" data-title="OpenAI open weight models now available on AWS" data-url="https://aws.amazon.com/blogs/aws/openai-open-weight-models-now-available-on-aws/&quot;&gt;&lt;p data-failed-message="Comments cannot be loaded… Please refresh and try again.">Loading comments…</p></div></aside>