diff --git a/changelog/september2024/2024-09-18-managed-inference-added-json-mode-and-structured-output.mdx b/changelog/september2024/2024-09-18-managed-inference-added-json-mode-and-structured-output.mdx
new file mode 100644
index 0000000000..a42a37997b
--- /dev/null
+++ b/changelog/september2024/2024-09-18-managed-inference-added-json-mode-and-structured-output.mdx
@@ -0,0 +1,15 @@
+---
+title: JSON mode and structured outputs
+status: added
+author:
+    fullname: 'Join the #ai channel on Slack.'
+    url: 'https://slack.scaleway.com'
+date: 2024-09-18
+category: ai-data
+product: managed-inference
+---
+
+All AI models can now reliably generate JSON output when required. The [Chat Completions API](https://www.scaleway.com/en/docs/ai-data/managed-inference/reference-content/openai-compatibility/#supported-parameters) supports the `response_format` parameter, which can be set to either `json_object` or `json_schema`, following OpenAI's specifications exactly. 
+
+Refer to [our documentation for code examples](https://www.scaleway.com/en/docs/ai-data/generative-apis/how-to/use-structured-outputs/), including usage with Pydantic models or manual schema definitions.
+