## Python SDK Changes:

* `glean.chat.create()`: 
  *  `request.text` **Added**
  *  `response.output[].content[].structured_output` **Added**

Author: speakeasybot

.speakeasy/gen.lock

@@ -34,7 +34,7 @@ generation:
     generateNewTests: true
     skipResponseBodyAssertions: true
 python:
-  version: 0.17.14
+  version: 0.17.15
   additionalDependencies:
     dev: {}
     main: {}

.speakeasy/gen.yaml

@@ -2,8 +2,8 @@ openapi: 3.1.0
 info:
   version: "0.9.0"
   title: Glean API
-  x-source-commit-sha: 5f1d873383438dc574c1075f9476915610e56875
-  x-open-api-commit-sha: 6a8c8e16bbbca13a130766a72f307fb373b8256a
+  x-source-commit-sha: cc70da4b874505308a45e85f609fa18587067cbf
+  x-open-api-commit-sha: 6446870bcd3126d795a07eedf83f09f10bdb52fa
   description: |
     # Introduction
     In addition to the data sources that Glean has built-in support for, Glean also provides a REST API that enables customers to put arbitrary content in the search index. This is useful, for example, for doing permissions-aware search over content in internal tools that reside on-prem as well as for searching over applications that Glean does not currently support first class. In addition these APIs allow the customer to push organization data (people info, organization structure etc) into Glean.
@@ -7444,13 +7444,72 @@ components:
         content:
           type: string
           minLength: 1
+    PlatformChatTextFormat:
+      type: object
+      additionalProperties: false
+      description: Unconstrained plain-text assistant output.
+      required:
+        - type
+      properties:
+        type:
+          type: string
+          description: Discriminator. Must be TEXT.
+          enum:
+            - TEXT
+    PlatformChatJsonSchemaFormat:
+      type: object
+      additionalProperties: false
+      description: Constrains the assistant response to a caller-supplied JSON Schema.
+      required:
+        - type
+        - name
+        - schema
+      properties:
+        type:
+          type: string
+          description: Discriminator. Must be JSON_SCHEMA.
+          enum:
+            - JSON_SCHEMA
+        name:
+          type: string
+          minLength: 1
+          maxLength: 64
+          description: A stable identifier for the schema.
+        description:
+          type: string
+          maxLength: 1024
+          description: An optional human-readable description of the schema.
+        schema:
+          type: object
+          additionalProperties: true
+          description: A JSON Schema object describing the desired output shape.
+        strict:
+          type: boolean
+          default: true
+          description: |
+            When true, the provider validates the response against the schema. Unsupported by some providers; defaults to true.
     PlatformChatCreateRequest:
       type: object
       additionalProperties: false
       required:
         - input
       example:
-        input: What is our parental leave policy?
+        input: Summarize our parental leave policy as JSON.
+        text:
+          format:
+            type: JSON_SCHEMA
+            name: policy_summary
+            schema:
+              type: object
+              properties:
+                eligible_employees:
+                  type: string
+                duration_weeks:
+                  type: integer
+              required:
+                - eligible_employees
+                - duration_weeks
+            strict: true
       properties:
         input:
           description: |
@@ -7477,6 +7536,18 @@ components:
           minLength: 1
           description: |
             Continue an existing stored conversation. Incompatible with message-array `input` and with `store: false`.
+        text:
+          type: object
+          additionalProperties: false
+          description: |
+            Optional configuration for the assistant's text response. When `format.type` is `JSON_SCHEMA`, the response is constrained to the supplied JSON schema and returned in `output[*].content[*].structured_output`. Structured output is not supported when `stream` is true.
+          properties:
+            format:
+              description: |
+                Output format for the assistant text. TEXT is unconstrained. JSON_SCHEMA constrains the response to the supplied schema.
+              oneOf:
+                - $ref: "#/components/schemas/PlatformChatTextFormat"
+                - $ref: "#/components/schemas/PlatformChatJsonSchemaFormat"
     PlatformChatStreamResponseId:
       type: string
       minLength: 1
@@ -7496,6 +7567,12 @@ components:
             - OUTPUT_TEXT
         text:
           type: string
+        structured_output:
+          type:
+            - object
+            - "null"
+          description: |
+            Parsed and validated JSON object when structured output was requested. Present only when the request included `text.format.type: JSON_SCHEMA`.
         annotations:
           type: array
           items:

.speakeasy/glean-merged-spec.yaml

@@ -159049,9 +159049,24 @@ workflows:
           requestBody:
             contentType: application/json
             payload:
-                input: What is our parental leave policy?
+                input: Summarize our parental leave policy as JSON.
                 store: true
                 stream: false
+                text:
+                    format:
+                        name: policy_summary
+                        schema:
+                            properties:
+                                duration_weeks:
+                                    type: integer
+                                eligible_employees:
+                                    type: string
+                            required:
+                                - eligible_employees
+                                - duration_weeks
+                            type: object
+                        strict: true
+                        type: JSON_SCHEMA
           successCriteria:
             - condition: $statusCode == 200
             - condition: $response.header.Content-Type == application/json

.speakeasy/tests.arazzo.yaml

@@ -2,8 +2,8 @@ speakeasyVersion: 1.797.0
 sources:
     Glean API:
         sourceNamespace: glean-api-specs
-        sourceRevisionDigest: sha256:1f470e422d6b202bf6bb80cc2e4b73077b7dab6d476cda804245fc0402ea42f1
-        sourceBlobDigest: sha256:8919e7ffe50c1ca78844cd89c216e8ab3c54a8e0003bd5072e7fe3cd35fb192c
+        sourceRevisionDigest: sha256:27df68e6ca0bfa00c188fa0e7056d3e56209d4ecdb4d0110c8687cf8532f4f3b
+        sourceBlobDigest: sha256:50b8e04438973d0945752232cfe3c4b685ed7e29e8107ba11791677020b49e64
         tags:
             - latest
     Glean Client API:
@@ -16,10 +16,10 @@ targets:
     glean:
         source: Glean API
         sourceNamespace: glean-api-specs
-        sourceRevisionDigest: sha256:1f470e422d6b202bf6bb80cc2e4b73077b7dab6d476cda804245fc0402ea42f1
-        sourceBlobDigest: sha256:8919e7ffe50c1ca78844cd89c216e8ab3c54a8e0003bd5072e7fe3cd35fb192c
+        sourceRevisionDigest: sha256:27df68e6ca0bfa00c188fa0e7056d3e56209d4ecdb4d0110c8687cf8532f4f3b
+        sourceBlobDigest: sha256:50b8e04438973d0945752232cfe3c4b685ed7e29e8107ba11791677020b49e64
         codeSamplesNamespace: glean-api-specs-python-code-samples
-        codeSamplesRevisionDigest: sha256:07e63ad12522c0ccf6dc9dab0943696972a489679ebc077bb6bfd23d25a123b8
+        codeSamplesRevisionDigest: sha256:83161070dcc75aa3b71b2fecd34814f1663270067089bd1e41206524697018c5
 workflow:
     workflowVersion: 1.0.0
     speakeasyVersion: latest

.speakeasy/workflow.lock

@@ -605,15 +605,36 @@ The stream is also a [Context Manager][context-manager] and can be used with the
 underlying connection when the context is exited.
 
 ```python
-from glean.api_client import Glean
+from glean.api_client import Glean, models
 import os
 
 
 with Glean(
     api_token=os.getenv("GLEAN_API_TOKEN", ""),
 ) as glean:
 
-    res = glean.chat.create_stream(input="What is our parental leave policy?", store=True)
+    res = glean.chat.create_stream(input="Summarize our parental leave policy as JSON.", store=True, text={
+        "format_": {
+            "type": models.PlatformChatJSONSchemaFormatType.JSON_SCHEMA,
+            "name": "policy_summary",
+            "schema_": {
+                "type": "object",
+                "properties": {
+                    "eligible_employees": {
+                        "type": "string",
+                    },
+                    "duration_weeks": {
+                        "type": "integer",
+                    },
+                },
+                "required": [
+                    "eligible_employees",
+                    "duration_weeks",
+                ],
+            },
+            "strict": True,
+        },
+    })
 
     with res as event_stream:
         for event in event_stream:

README.md

@@ -998,4 +998,14 @@ Based on:
 ### Generated
 - [python v0.17.14] .
 ### Releases
-- [PyPI v0.17.14] https://pypi.org/project/glean-api-client/0.17.14 - .
+- [PyPI v0.17.14] https://pypi.org/project/glean-api-client/0.17.14 - .
+
+## 2026-09-18 17:09:14
+### Changes
+Based on:
+- OpenAPI Doc  
+- Speakeasy CLI 1.797.0 (2.937.18) https://github.com/speakeasy-api/speakeasy
+### Generated
+- [python v0.17.15] .
+### Releases
+- [PyPI v0.17.15] https://pypi.org/project/glean-api-client/0.17.15 - .

RELEASES.md

@@ -0,0 +1,20 @@
+# PlatformChatCreateFormat
+
+Output format for the assistant text. TEXT is unconstrained. JSON_SCHEMA constrains the response to the supplied schema.
+
+
+
+## Supported Types
+
+### `models.PlatformChatTextFormat`
+
+```python
+value: models.PlatformChatTextFormat = /* values here */
+```
+
+### `models.PlatformChatJSONSchemaFormat`
+
+```python
+value: models.PlatformChatJSONSchemaFormat = /* values here */
+```
+