AWS bedrock documentation change

Service: bedrock · 2025-11-22 · Documentation low

File: bedrock/latest/userguide/custom-model-import-code-samples.md

Summary

Added architecture support details and API limitations for GPT-OSS models

Security assessment

Changes clarify API compatibility requirements but do not address security vulnerabilities or introduce security features. Focuses on functional limitations rather than security controls.

Diff

diff --git a/bedrock/latest/userguide/custom-model-import-code-samples.md b/bedrock/latest/userguide/custom-model-import-code-samples.md
index 2722d8e60..8534e6020 100644
--- a//bedrock/latest/userguide/custom-model-import-code-samples.md
+++ b//bedrock/latest/userguide/custom-model-import-code-samples.md
@@ -4,0 +5,2 @@
+Important: Model Architecture Support
+
@@ -7 +9,24 @@
-The following code samples show how to set up permissions, create a custom model import job, view the details of your import jobs and imported models, and delete imported model.
+The following code samples show how to set up permissions, create a custom model import job, view the details of your import jobs and imported models, and delete imported model. These code samples apply to Mistral AI, Llama, Qwen, Flan, GPTBigCode, and Mixtral architectures.
+
+## Important: Model Architecture Support
+
+**GPT-OSS Model Limitations:**
+
+  * **Converse API Not Supported:** GPT-OSS based custom model import models do NOT support the Converse API or ConverseStream API.
+
+  * **Use InvokeModel API Instead:** Customers must use the [InvokeModel](https://docs.aws.amazon.com/bedrock/latest/APIReference/API_runtime_InvokeModel.html) API when working with GPT-OSS based custom models.
+
+  * **API Schema Requirement:** GPT-OSS models require OpenAI-compatible API schemas:
+
+    * Completion format for completion requests
+
+    * ChatCompletion format for chat requests
+
+    * Response format follows OpenAI API specifications
+
+  * **Supported Models for Converse API:** The Converse API is supported for Mistral AI, Llama, Qwen, Flan, GPTBigCode, and Mixtral architectures only.
+
+
+
+
+For GPT-OSS model usage examples with [InvokeModel](https://docs.aws.amazon.com/bedrock/latest/APIReference/API_runtime_InvokeModel.html) API, refer to the OpenAI API documentation for request/response schemas.