diff --git a/custom-words.txt b/custom-words.txt
index c2c806663e3e..853419ac77a5 100644
--- a/custom-words.txt
+++ b/custom-words.txt
@@ -967,6 +967,7 @@ findsquare
 finetune
 finetunes
 firmwares
+flac
 flexera
 flink
 flowlet
@@ -1644,6 +1645,7 @@ moodle
 mooncake
 mountainview
 movere
+mpga
 mpls
 mpns
 mran
diff --git a/specification/cognitiveservices/OpenAI.Inference/examples/2023-09-01-preview/audio_transcription_object.json b/specification/cognitiveservices/OpenAI.Inference/examples/2023-09-01-preview/audio_transcription_object.json
index dc8c33a10a17..14fa046829a8 100644
--- a/specification/cognitiveservices/OpenAI.Inference/examples/2023-09-01-preview/audio_transcription_object.json
+++ b/specification/cognitiveservices/OpenAI.Inference/examples/2023-09-01-preview/audio_transcription_object.json
@@ -4,7 +4,8 @@
   "parameters": {
     "endpoint": "{endpoint}",
     "api-version": "2023-09-01-preview",
-    "deploymentId": "<deployment-id>"
+    "deploymentId": "<deployment-id>",
+    "body": "---multipart-boundary\nContent-Disposition: form-data; name=\"file\"; filename=\"file.wav\"\nContent-Type: application/octet-stream\n\nRIFF..audio.data.omitted\n---multipart-boundary--"
   },
   "responses": {
     "200": {
diff --git a/specification/cognitiveservices/OpenAI.Inference/main.tsp b/specification/cognitiveservices/OpenAI.Inference/main.tsp
index 9f58a864dad9..3249779f8486 100644
--- a/specification/cognitiveservices/OpenAI.Inference/main.tsp
+++ b/specification/cognitiveservices/OpenAI.Inference/main.tsp
@@ -1,11 +1,14 @@
-import "@typespec/rest";
 import "@typespec/http";
 import "@typespec/versioning";
 
-import "./routes.tsp";
+import "./routes/common.tsp";
+import "./routes/audio_transcription.tsp";
+import "./routes/audio_translation.tsp";
+import "./routes/completions.tsp";
+import "./routes/images.tsp";
+import "./routes/embeddings.tsp";
 
 using TypeSpec.Http;
-using TypeSpec.Rest;
 using TypeSpec.Versioning;
 
 #suppress "@azure-tools/typespec-azure-core/casing-style" "OpenAI is a case-sensitive name"
@@ -56,13 +59,3 @@ enum ServiceApiVersions {
   @useDependency(Azure.Core.Versions.v1_0_Preview_1)
   v2023_09_01_Preview: "2023-09-01-preview",
 }
-
-@doc("A specific deployment")
-@TypeSpec.Rest.resource("deployments")
-model Deployment {
-  @visibility("read")
-  @doc("Specifies either the model deployment name (when using Azure OpenAI) or model name (when using non-Azure OpenAI) to use for this request.")
-  @projectedName("java", "deploymentOrModelName")
-  @key
-  deploymentId: string;
-}
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/audio.transcription.tsp b/specification/cognitiveservices/OpenAI.Inference/models/audio/audio_transcription.tsp
similarity index 98%
rename from specification/cognitiveservices/OpenAI.Inference/models/audio.transcription.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/audio/audio_transcription.tsp
index 9a1af2588457..aea912a57508 100644
--- a/specification/cognitiveservices/OpenAI.Inference/models/audio.transcription.tsp
+++ b/specification/cognitiveservices/OpenAI.Inference/models/audio/audio_transcription.tsp
@@ -1,5 +1,5 @@
 import "@typespec/versioning";
-import "./audio.common.tsp";
+import "./common.tsp";
 
 namespace Azure.OpenAI;
 
@@ -36,7 +36,7 @@ The configuration information for an audio transcription request.
 model AudioTranscriptionOptions {
   @doc("""
   The audio data to transcribe. This must be the binary content of a file in one of the supported media formats:
-  flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm.
+   flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm.
   """)
   @projectedName("csharp", "AudioData")
   file: bytes;
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/audio.translation.tsp b/specification/cognitiveservices/OpenAI.Inference/models/audio/audio_translation.tsp
similarity index 98%
rename from specification/cognitiveservices/OpenAI.Inference/models/audio.translation.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/audio/audio_translation.tsp
index 2d40e3fb5cb2..1638107caf8a 100644
--- a/specification/cognitiveservices/OpenAI.Inference/models/audio.translation.tsp
+++ b/specification/cognitiveservices/OpenAI.Inference/models/audio/audio_translation.tsp
@@ -1,5 +1,5 @@
 import "@typespec/versioning";
-import "./audio.common.tsp";
+import "./common.tsp";
 
 namespace Azure.OpenAI;
 
@@ -36,7 +36,7 @@ The configuration information for an audio translation request.
 model AudioTranslationOptions {
   @doc("""
   The audio data to translate. This must be the binary content of a file in one of the supported media formats:
-  flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm.
+   flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm.
   """)
   @projectedName("csharp", "AudioData")
   file: bytes;
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/audio.common.tsp b/specification/cognitiveservices/OpenAI.Inference/models/audio/common.tsp
similarity index 100%
rename from specification/cognitiveservices/OpenAI.Inference/models/audio.common.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/audio/common.tsp
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/azure.chat.extensions.tsp b/specification/cognitiveservices/OpenAI.Inference/models/completions/azure_chat_extensions.tsp
similarity index 97%
rename from specification/cognitiveservices/OpenAI.Inference/models/azure.chat.extensions.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/completions/azure_chat_extensions.tsp
index 18bd55d7421e..9f024f1f0312 100644
--- a/specification/cognitiveservices/OpenAI.Inference/models/azure.chat.extensions.tsp
+++ b/specification/cognitiveservices/OpenAI.Inference/models/completions/azure_chat_extensions.tsp
@@ -1,6 +1,7 @@
 import "@typespec/versioning";
-import "./chat.completions.tsp";
-import "./azure.cognitive.search.options.tsp";
+
+import "./chat_completions.tsp";
+import "./azure_cognitive_search_options.tsp";
 
 namespace Azure.OpenAI;
 
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/azure.cognitive.search.options.tsp b/specification/cognitiveservices/OpenAI.Inference/models/completions/azure_cognitive_search_options.tsp
similarity index 98%
rename from specification/cognitiveservices/OpenAI.Inference/models/azure.cognitive.search.options.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/completions/azure_cognitive_search_options.tsp
index 2f7e0f878fbb..eebe186dc478 100644
--- a/specification/cognitiveservices/OpenAI.Inference/models/azure.cognitive.search.options.tsp
+++ b/specification/cognitiveservices/OpenAI.Inference/models/completions/azure_cognitive_search_options.tsp
@@ -1,5 +1,6 @@
 import "@typespec/versioning";
-import "./azure.chat.extensions.tsp";
+
+import "./azure_chat_extensions.tsp";
 
 namespace Azure.OpenAI;
 
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/chat.completions.tsp b/specification/cognitiveservices/OpenAI.Inference/models/completions/chat_completions.tsp
similarity index 98%
rename from specification/cognitiveservices/OpenAI.Inference/models/chat.completions.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/completions/chat_completions.tsp
index de754b192814..aad348dcc19c 100644
--- a/specification/cognitiveservices/OpenAI.Inference/models/chat.completions.tsp
+++ b/specification/cognitiveservices/OpenAI.Inference/models/completions/chat_completions.tsp
@@ -1,8 +1,9 @@
 import "@typespec/rest";
 import "@typespec/http";
 import "@typespec/versioning";
-import "./azure.chat.extensions.tsp";
-import "./completions.common.tsp";
+
+import "./azure_chat_extensions.tsp";
+import "./common.tsp";
 import "./functions.tsp";
 
 using TypeSpec.Rest;
@@ -231,8 +232,8 @@ model ChatChoice {
   delta?: ChatMessage;
 
   @doc("""
-    Information about the content filtering category (hate, sexual, violence, self_harm), if it 
-    has been detected, as well as the severity level (very_low, low, medium, high-scale that 
+    Information about the content filtering category (hate, sexual, violence, self_harm), if it
+    has been detected, as well as the severity level (very_low, low, medium, high-scale that
     determines the intensity and risk level of harmful content) and if it has been filtered or not.
     """)
   @added(ServiceApiVersions.v2023_06_01_Preview)
@@ -268,7 +269,7 @@ model ChatCompletions {
   choices: ChatChoice[];
 
   @doc("""
-  Content filtering results for zero or more prompts in the request. In a streaming request, 
+  Content filtering results for zero or more prompts in the request. In a streaming request,
   results for different prompts may arrive at different times or in different orders.
   """)
   @added(ServiceApiVersions.v2023_06_01_Preview)
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/completions.common.tsp b/specification/cognitiveservices/OpenAI.Inference/models/completions/common.tsp
similarity index 100%
rename from specification/cognitiveservices/OpenAI.Inference/models/completions.common.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/completions/common.tsp
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/completions.create.tsp b/specification/cognitiveservices/OpenAI.Inference/models/completions/completions_create.tsp
similarity index 98%
rename from specification/cognitiveservices/OpenAI.Inference/models/completions.create.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/completions/completions_create.tsp
index a4f73016e4db..93dcb25733a7 100644
--- a/specification/cognitiveservices/OpenAI.Inference/models/completions.create.tsp
+++ b/specification/cognitiveservices/OpenAI.Inference/models/completions/completions_create.tsp
@@ -1,7 +1,8 @@
 import "@typespec/rest";
 import "@typespec/http";
 import "@typespec/versioning";
-import "./completions.common.tsp";
+
+import "./common.tsp";
 
 using TypeSpec.Rest;
 using TypeSpec.Http;
@@ -163,7 +164,7 @@ model Completions {
   created: utcDateTime;
 
   @doc("""
-  Content filtering results for zero or more prompts in the request. In a streaming request, 
+  Content filtering results for zero or more prompts in the request. In a streaming request,
   results for different prompts may arrive at different times or in different orders.
   """)
   @added(ServiceApiVersions.v2023_06_01_Preview)
@@ -200,8 +201,8 @@ model Choice {
   index: int32;
 
   @doc("""
-  Information about the content filtering category (hate, sexual, violence, self_harm), if it 
-  has been detected, as well as the severity level (very_low, low, medium, high-scale that 
+  Information about the content filtering category (hate, sexual, violence, self_harm), if it
+  has been detected, as well as the severity level (very_low, low, medium, high-scale that
   determines the intensity and risk level of harmful content) and if it has been filtered or not.
   """)
   @added(ServiceApiVersions.v2023_06_01_Preview)
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/functions.tsp b/specification/cognitiveservices/OpenAI.Inference/models/completions/functions.tsp
similarity index 100%
rename from specification/cognitiveservices/OpenAI.Inference/models/functions.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/completions/functions.tsp
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/embeddings.create.tsp b/specification/cognitiveservices/OpenAI.Inference/models/embeddings_create.tsp
similarity index 100%
rename from specification/cognitiveservices/OpenAI.Inference/models/embeddings.create.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/embeddings_create.tsp
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/operations.common.tsp b/specification/cognitiveservices/OpenAI.Inference/models/images/common.tsp
similarity index 100%
rename from specification/cognitiveservices/OpenAI.Inference/models/operations.common.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/images/common.tsp
diff --git a/specification/cognitiveservices/OpenAI.Inference/models/images.tsp b/specification/cognitiveservices/OpenAI.Inference/models/images/images.tsp
similarity index 99%
rename from specification/cognitiveservices/OpenAI.Inference/models/images.tsp
rename to specification/cognitiveservices/OpenAI.Inference/models/images/images.tsp
index d2d8530fe5e1..7542dce12326 100644
--- a/specification/cognitiveservices/OpenAI.Inference/models/images.tsp
+++ b/specification/cognitiveservices/OpenAI.Inference/models/images/images.tsp
@@ -1,7 +1,8 @@
 import "@azure-tools/typespec-azure-core";
 import "@typespec/http";
 import "@typespec/versioning";
-import "./operations.common.tsp";
+
+import "./common.tsp";
 
 using Azure.Core;
 using TypeSpec.Http;
diff --git a/specification/cognitiveservices/OpenAI.Inference/routes.tsp b/specification/cognitiveservices/OpenAI.Inference/routes.tsp
deleted file mode 100644
index a19add16c825..000000000000
--- a/specification/cognitiveservices/OpenAI.Inference/routes.tsp
+++ /dev/null
@@ -1,180 +0,0 @@
-import "@azure-tools/typespec-azure-core";
-import "@typespec/rest";
-import "@typespec/http";
-import "@typespec/versioning";
-
-import "./models/completions.create.tsp";
-import "./models/chat.completions.tsp";
-import "./models/embeddings.create.tsp";
-import "./models/images.tsp";
-import "./models/audio.transcription.tsp";
-import "./models/audio.translation.tsp";
-
-using TypeSpec.Rest;
-using TypeSpec.Http;
-using TypeSpec.Versioning;
-
-using Azure.Core;
-
-namespace Azure.OpenAI;
-
-@doc("Return the embeddings for a given prompt.")
-@actionSeparator("/")
-@action("embeddings") //@convenientAPI(true)
-op getEmbeddings is Azure.Core.ResourceAction<
-  Deployment,
-  EmbeddingsOptions,
-  Embeddings
->;
-
-@doc("""
-Gets completions for the provided input prompts.
-Completions support a wide variety of tasks and generate text that continues from or "completes"
-provided prompt data.
-""")
-@actionSeparator("/")
-@action("completions") //@convenientAPI(true)
-op getCompletions is Azure.Core.ResourceAction<
-  Deployment,
-  CompletionsOptions,
-  Completions
->;
-
-@doc("""
-Gets chat completions for the provided chat messages.
-Completions support a wide variety of tasks and generate text that continues from or "completes"
-provided prompt data.
-""")
-@added(ServiceApiVersions.v2023_05_15)
-@actionSeparator("/")
-@action("chat/completions") //@convenientAPI(true)
-op getChatCompletions is ResourceAction<
-  Deployment,
-  ChatCompletionsOptions,
-  ChatCompletions
->;
-
-@doc("""
-Gets chat completions for the provided chat messages.
-This is an Azure-specific version of chat completions that supports integration with configured data sources and
-other augmentations to the base chat completions capabilities.
-""")
-// Note: client libraries should abstract the existence of this as a distinct REST route and instead perform dynamic
-//        selection of the appropriate request URL structure based on the presence or non-presence of matching
-//        chat extensions configuration in the request options model. Future service API versions may merge this
-//        capability into the standard /chat/completions route.
-@added(ServiceApiVersions.v2023_08_01_Preview)
-@actionSeparator("/")
-@action("extensions/chat/completions")
-op getChatCompletionsWithAzureExtensions is ResourceAction<
-  Deployment,
-  ChatCompletionsOptions,
-  ChatCompletions
->;
-
-// Note: pending resolution of  cross-language code emission behavior for long-running operations, image generation
-//        reuses its final operation response model as its status polling model.
-#suppress "@azure-tools/typespec-azure-core/rpc-operation-request-body" "MUST fix in next update"
-#suppress "@azure-tools/typespec-azure-core/no-rpc-path-params" "Allowed because this is a non-standard status polling operation."
-@doc("Returns the status of the images operation")
-@added(ServiceApiVersions.v2023_06_01_Preview)
-@route("/operations/images/{operationId}")
-op getAzureBatchImageGenerationOperationStatus is RpcOperation<
-  {
-    @doc(".") @path operationId: string;
-  },
-  BatchImageGenerationOperationResponse
->;
-
-#suppress "@azure-tools/typespec-azure-core/use-standard-operations" ""
-@doc("Starts the generation of a batch of images from a text caption")
-@added(ServiceApiVersions.v2023_06_01_Preview)
-@route("/images/generations:submit")
-@pollingOperation(
-  getAzureBatchImageGenerationOperationStatus,
-  {
-    operationId: ResponseProperty<"id">,
-  }
-)
-op beginAzureBatchImageGeneration is OaiLongRunningRpcOperation<
-  ImageGenerationOptions,
-  BatchImageGenerationOperationResponse,
-  BatchImageGenerationOperationResponse
->;
-
-// Audio implementation note: depending on the response_format provided in operation request bodies, these operations
-// may respond with a response body that's either a JSON object or a text/plain string. Pending more precise
-// specification of the relationship between the input format selection and resulting output response format, the two
-// content-type-specific options are provided for each operation with no delineation between the matching request body
-// requirements. These content-type-specific operations should be hidden by clients and combined into single operations
-// that share common characteristics whenever possible.
-
-alias MultipartFormDataRequestHeadersTraits = Azure.Core.Traits.RequestHeadersTrait<{
-  @doc("The content type for the operation. Always multipart/form-data for this operation.")
-  @header("content-type")
-  contentType: "multipart/form-data";
-}>;
-
-#suppress "@azure-tools/typespec-azure-core/byos" "representation of existing multipart/form-data operation"
-#suppress "@azure-tools/typespec-azure-core/use-standard-operations" "non-model response value needed"
-@doc("""
-Gets transcribed text and associated metadata from provided spoken audio data. Audio will be transcribed in the
-written language corresponding to the language it was spoken in.
-""")
-@added(ServiceApiVersions.v2023_09_01_Preview)
-@actionSeparator("/")
-@action("audio/transcriptions")
-@sharedRoute
-op getAudioTranscriptionAsPlainText is Azure.Core.Foundations.ResourceOperation<
-  Deployment,
-  AudioTranscriptionOptions,  // response_format must be one of: text, srt, vtt
-  string,
-  MultipartFormDataRequestHeadersTraits
->;
-
-#suppress "@azure-tools/typespec-azure-core/byos" "representation of existing multipart/form-data operation"
-@doc("""
-Gets transcribed text and associated metadata from provided spoken audio data. Audio will be transcribed in the
-written language corresponding to the language it was spoken in.
-""")
-@added(ServiceApiVersions.v2023_09_01_Preview)
-@actionSeparator("/")
-@action("audio/transcriptions")
-@sharedRoute
-op getAudioTranscriptionAsResponseObject is Azure.Core.ResourceAction<
-  Deployment,
-  AudioTranscriptionOptions,  // response_format must be unspecified (json) or one of: json, verbose_json
-  AudioTranscription,
-  MultipartFormDataRequestHeadersTraits
->;
-
-#suppress "@azure-tools/typespec-azure-core/byos" "representation of existing multipart/form-data operation"
-#suppress "@azure-tools/typespec-azure-core/use-standard-operations" "non-model response value needed"
-@doc("""
-Gets English language transcribed text and associated metadata from provided spoken audio data.
-""")
-@added(ServiceApiVersions.v2023_09_01_Preview)
-@actionSeparator("/")
-@action("audio/translations")
-@sharedRoute
-op getAudioTranslationAsPlainText is Azure.Core.Foundations.ResourceOperation<
-  Deployment,
-  AudioTranslationOptions,  // response_format must be one of: text, srt, vtt
-  string,
-  MultipartFormDataRequestHeadersTraits
->;
-
-#suppress "@azure-tools/typespec-azure-core/byos" "representation of existing multipart/form-data operation"
-@doc("""
-Gets English language transcribed text and associated metadata from provided spoken audio data.
-""")
-@added(ServiceApiVersions.v2023_09_01_Preview)
-@actionSeparator("/")
-@action("audio/translations")
-@sharedRoute
-op getAudioTranslationAsResponseObject is Azure.Core.ResourceAction<
-  Deployment,
-  AudioTranslationOptions,  // response_format must be unspecified (json) or one of: json, verbose_json
-  AudioTranslation,
-  MultipartFormDataRequestHeadersTraits
->;
diff --git a/specification/cognitiveservices/OpenAI.Inference/routes/audio_transcription.tsp b/specification/cognitiveservices/OpenAI.Inference/routes/audio_transcription.tsp
new file mode 100644
index 000000000000..3a4565c36ba3
--- /dev/null
+++ b/specification/cognitiveservices/OpenAI.Inference/routes/audio_transcription.tsp
@@ -0,0 +1,50 @@
+import "@azure-tools/typespec-azure-core";
+import "@typespec/rest";
+import "@typespec/http";
+import "@typespec/versioning";
+
+import "../models/audio/audio_transcription.tsp";
+import "./common.tsp";
+
+using TypeSpec.Rest;
+using TypeSpec.Http;
+using TypeSpec.Versioning;
+
+using Azure.Core;
+
+namespace Azure.OpenAI;
+
+// The order in which the operations are defined matters for the purpose of OpenAPIv2 generation
+
+#suppress "@azure-tools/typespec-azure-core/byos" "representation of existing multipart/form-data operation"
+#suppress "@azure-tools/typespec-azure-core/use-standard-operations" "non-model response value needed"
+@doc("""
+Gets transcribed text and associated metadata from provided spoken audio data. Audio will be transcribed in the
+written language corresponding to the language it was spoken in.
+""")
+@added(ServiceApiVersions.v2023_09_01_Preview)
+@actionSeparator("/")
+@action("audio/transcriptions")
+@sharedRoute
+op getAudioTranscriptionAsPlainText is Azure.Core.Foundations.ResourceOperation<
+    Deployment,
+    AudioTranscriptionOptions,  // response_format must be one of: text, srt, vtt
+    string,
+    MultipartFormDataRequestHeadersTraits
+>;
+
+#suppress "@azure-tools/typespec-azure-core/byos" "representation of existing multipart/form-data operation"
+@doc("""
+Gets transcribed text and associated metadata from provided spoken audio data. Audio will be transcribed in the
+written language corresponding to the language it was spoken in.
+""")
+@added(ServiceApiVersions.v2023_09_01_Preview)
+@actionSeparator("/")
+@action("audio/transcriptions")
+@sharedRoute
+op getAudioTranscriptionAsResponseObject is Azure.Core.ResourceAction<
+    Deployment,
+    AudioTranscriptionOptions,  // response_format must be unspecified (json) or one of: json, verbose_json
+    AudioTranscription,
+    MultipartFormDataRequestHeadersTraits
+>;
diff --git a/specification/cognitiveservices/OpenAI.Inference/routes/audio_translation.tsp b/specification/cognitiveservices/OpenAI.Inference/routes/audio_translation.tsp
new file mode 100644
index 000000000000..fb378350e549
--- /dev/null
+++ b/specification/cognitiveservices/OpenAI.Inference/routes/audio_translation.tsp
@@ -0,0 +1,48 @@
+import "@azure-tools/typespec-azure-core";
+import "@typespec/rest";
+import "@typespec/http";
+import "@typespec/versioning";
+
+import "../models/audio/audio_translation.tsp";
+import "./common.tsp";
+
+using TypeSpec.Rest;
+using TypeSpec.Http;
+using TypeSpec.Versioning;
+
+using Azure.Core;
+
+namespace Azure.OpenAI;
+
+// The order in which the operations are defined matters for the purpose of OpenAPIv2 generation
+
+#suppress "@azure-tools/typespec-azure-core/byos" "representation of existing multipart/form-data operation"
+#suppress "@azure-tools/typespec-azure-core/use-standard-operations" "non-model response value needed"
+@doc("""
+Gets English language transcribed text and associated metadata from provided spoken audio data.
+""")
+@added(ServiceApiVersions.v2023_09_01_Preview)
+@actionSeparator("/")
+@action("audio/translations")
+@sharedRoute
+op getAudioTranslationAsPlainText is Azure.Core.Foundations.ResourceOperation<
+    Deployment,
+    AudioTranslationOptions,  // response_format must be one of: text, srt, vtt
+    string,
+    MultipartFormDataRequestHeadersTraits
+>;
+
+#suppress "@azure-tools/typespec-azure-core/byos" "representation of existing multipart/form-data operation"
+@doc("""
+Gets English language transcribed text and associated metadata from provided spoken audio data.
+""")
+@added(ServiceApiVersions.v2023_09_01_Preview)
+@actionSeparator("/")
+@action("audio/translations")
+@sharedRoute
+op getAudioTranslationAsResponseObject is Azure.Core.ResourceAction<
+    Deployment,
+    AudioTranslationOptions,  // response_format must be unspecified (json) or one of: json, verbose_json
+    AudioTranslation,
+    MultipartFormDataRequestHeadersTraits
+>;
diff --git a/specification/cognitiveservices/OpenAI.Inference/routes/common.tsp b/specification/cognitiveservices/OpenAI.Inference/routes/common.tsp
new file mode 100644
index 000000000000..33b175d380dc
--- /dev/null
+++ b/specification/cognitiveservices/OpenAI.Inference/routes/common.tsp
@@ -0,0 +1,32 @@
+import "@typespec/rest";
+import "@typespec/http";
+// import "@typespec/versioning";
+
+// import "./routes.tsp";
+
+using TypeSpec.Http;
+using TypeSpec.Rest;
+// using TypeSpec.Versioning;
+
+@doc("A specific deployment")
+@TypeSpec.Rest.resource("deployments")
+model Deployment {
+    @visibility("read")
+    @doc("Specifies either the model deployment name (when using Azure OpenAI) or model name (when using non-Azure OpenAI) to use for this request.")
+    @projectedName("java", "deploymentOrModelName")
+    @key
+    deploymentId: string;
+}
+
+// Audio implementation note: depending on the response_format provided in operation request bodies, these operations
+// may respond with a response body that's either a JSON object or a text/plain string. Pending more precise
+// specification of the relationship between the input format selection and resulting output response format, the two
+// content-type-specific options are provided for each operation with no delineation between the matching request body
+// requirements. These content-type-specific operations should be hidden by clients and combined into single operations
+// that share common characteristics whenever possible.
+
+alias MultipartFormDataRequestHeadersTraits = Azure.Core.Traits.RequestHeadersTrait<{
+    @doc("The content type for the operation. Always multipart/form-data for this operation.")
+    @header("content-type")
+    contentType: "multipart/form-data";
+}>;
diff --git a/specification/cognitiveservices/OpenAI.Inference/routes/completions.tsp b/specification/cognitiveservices/OpenAI.Inference/routes/completions.tsp
new file mode 100644
index 000000000000..41d538bbbb45
--- /dev/null
+++ b/specification/cognitiveservices/OpenAI.Inference/routes/completions.tsp
@@ -0,0 +1,60 @@
+import "@azure-tools/typespec-azure-core";
+import "@typespec/rest";
+import "@typespec/http";
+import "@typespec/versioning";
+
+import "../models/completions/completions_create.tsp";
+import "../models/completions/chat_completions.tsp";
+
+using TypeSpec.Rest;
+using TypeSpec.Http;
+using TypeSpec.Versioning;
+
+using Azure.Core;
+
+namespace Azure.OpenAI;
+
+@doc("""
+Gets completions for the provided input prompts.
+Completions support a wide variety of tasks and generate text that continues from or "completes"
+provided prompt data.
+""")
+@actionSeparator("/")
+@action("completions") //@convenientAPI(true)
+op getCompletions is Azure.Core.ResourceAction<
+    Deployment,
+    CompletionsOptions,
+    Completions
+>;
+
+@doc("""
+Gets chat completions for the provided chat messages.
+Completions support a wide variety of tasks and generate text that continues from or "completes"
+provided prompt data.
+""")
+@added(ServiceApiVersions.v2023_05_15)
+@actionSeparator("/")
+@action("chat/completions") //@convenientAPI(true)
+op getChatCompletions is ResourceAction<
+    Deployment,
+    ChatCompletionsOptions,
+    ChatCompletions
+>;
+
+@doc("""
+Gets chat completions for the provided chat messages.
+This is an Azure-specific version of chat completions that supports integration with configured data sources and
+other augmentations to the base chat completions capabilities.
+""")
+// Note: client libraries should abstract the existence of this as a distinct REST route and instead perform dynamic
+//        selection of the appropriate request URL structure based on the presence or non-presence of matching
+//        chat extensions configuration in the request options model. Future service API versions may merge this
+//        capability into the standard /chat/completions route.
+@added(ServiceApiVersions.v2023_08_01_Preview)
+@actionSeparator("/")
+@action("extensions/chat/completions")
+op getChatCompletionsWithAzureExtensions is ResourceAction<
+    Deployment,
+    ChatCompletionsOptions,
+    ChatCompletions
+>;
diff --git a/specification/cognitiveservices/OpenAI.Inference/routes/embeddings.tsp b/specification/cognitiveservices/OpenAI.Inference/routes/embeddings.tsp
new file mode 100644
index 000000000000..c2a6722bdaa4
--- /dev/null
+++ b/specification/cognitiveservices/OpenAI.Inference/routes/embeddings.tsp
@@ -0,0 +1,23 @@
+import "@azure-tools/typespec-azure-core";
+import "@typespec/rest";
+import "@typespec/http";
+import "@typespec/versioning";
+
+import "../models/embeddings_create.tsp";
+
+using TypeSpec.Rest;
+using TypeSpec.Http;
+using TypeSpec.Versioning;
+
+using Azure.Core;
+
+namespace Azure.OpenAI;
+
+@doc("Return the embeddings for a given prompt.")
+@actionSeparator("/")
+@action("embeddings") //@convenientAPI(true)
+op getEmbeddings is Azure.Core.ResourceAction<
+    Deployment,
+    EmbeddingsOptions,
+    Embeddings
+>;
diff --git a/specification/cognitiveservices/OpenAI.Inference/routes/images.tsp b/specification/cognitiveservices/OpenAI.Inference/routes/images.tsp
new file mode 100644
index 000000000000..c9dce14161df
--- /dev/null
+++ b/specification/cognitiveservices/OpenAI.Inference/routes/images.tsp
@@ -0,0 +1,46 @@
+import "@azure-tools/typespec-azure-core";
+import "@typespec/rest";
+import "@typespec/http";
+import "@typespec/versioning";
+
+import "../models/images/images.tsp";
+import "../models/images/common.tsp";
+
+using TypeSpec.Rest;
+using TypeSpec.Http;
+using TypeSpec.Versioning;
+
+using Azure.Core;
+
+namespace Azure.OpenAI;
+
+// Note: pending resolution of  cross-language code emission behavior for long-running operations, image generation
+// reuses its final operation response model as its status polling model.
+
+#suppress "@azure-tools/typespec-azure-core/rpc-operation-request-body" "MUST fix in next update"
+#suppress "@azure-tools/typespec-azure-core/no-rpc-path-params" "Allowed because this is a non-standard status polling operation."
+@doc("Returns the status of the images operation")
+@added(ServiceApiVersions.v2023_06_01_Preview)
+@route("/operations/images/{operationId}")
+op getAzureBatchImageGenerationOperationStatus is RpcOperation<
+    {
+        @doc(".") @path operationId: string;
+    },
+    BatchImageGenerationOperationResponse
+>;
+
+#suppress "@azure-tools/typespec-azure-core/use-standard-operations" ""
+@doc("Starts the generation of a batch of images from a text caption")
+@added(ServiceApiVersions.v2023_06_01_Preview)
+@route("/images/generations:submit")
+@pollingOperation(
+    getAzureBatchImageGenerationOperationStatus,
+    {
+        operationId: ResponseProperty<"id">,
+    }
+)
+op beginAzureBatchImageGeneration is OaiLongRunningRpcOperation<
+    ImageGenerationOptions,
+    BatchImageGenerationOperationResponse,
+    BatchImageGenerationOperationResponse
+>;
diff --git a/specification/cognitiveservices/OpenAI.Inference/tspconfig.yaml b/specification/cognitiveservices/OpenAI.Inference/tspconfig.yaml
index 4e2b82d3e928..50d80c70b83e 100644
--- a/specification/cognitiveservices/OpenAI.Inference/tspconfig.yaml
+++ b/specification/cognitiveservices/OpenAI.Inference/tspconfig.yaml
@@ -13,7 +13,7 @@ options:
     emitter-output-dir: "{project-root}/../"
     output-file: "{azure-resource-provider-folder}/AzureOpenAI/inference/{version-status}/{version}/generated.json"
     azure-resource-provider-folder: "data-plane"
-    examples-directory: examples
+    examples-directory: "{project-root}/examples"
     omit-unreachable-types: true
   "@azure-tools/typespec-csharp":
     package-dir: "Azure.AI.OpenAI"
diff --git a/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-06-01-preview/generated.json b/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-06-01-preview/generated.json
index 560081a2d09d..0a5686339dbe 100644
--- a/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-06-01-preview/generated.json
+++ b/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-06-01-preview/generated.json
@@ -474,7 +474,7 @@
         },
         "content_filter_results": {
           "$ref": "#/definitions/ContentFilterResults",
-          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it \nhas been detected, as well as the severity level (very_low, low, medium, high-scale that \ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
+          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it\nhas been detected, as well as the severity level (very_low, low, medium, high-scale that\ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
           "x-ms-client-name": "contentFilterResults"
         }
       },
@@ -506,7 +506,7 @@
         },
         "prompt_filter_results": {
           "type": "array",
-          "description": "Content filtering results for zero or more prompts in the request. In a streaming request, \nresults for different prompts may arrive at different times or in different orders.",
+          "description": "Content filtering results for zero or more prompts in the request. In a streaming request,\nresults for different prompts may arrive at different times or in different orders.",
           "items": {
             "$ref": "#/definitions/PromptFilterResult"
           },
@@ -668,7 +668,7 @@
         },
         "content_filter_results": {
           "$ref": "#/definitions/ContentFilterResults",
-          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it \nhas been detected, as well as the severity level (very_low, low, medium, high-scale that \ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
+          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it\nhas been detected, as well as the severity level (very_low, low, medium, high-scale that\ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
           "x-ms-client-name": "contentFilterResults"
         },
         "logprobs": {
@@ -710,7 +710,7 @@
         },
         "prompt_filter_results": {
           "type": "array",
-          "description": "Content filtering results for zero or more prompts in the request. In a streaming request, \nresults for different prompts may arrive at different times or in different orders.",
+          "description": "Content filtering results for zero or more prompts in the request. In a streaming request,\nresults for different prompts may arrive at different times or in different orders.",
           "items": {
             "$ref": "#/definitions/PromptFilterResult"
           },
diff --git a/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-07-01-preview/generated.json b/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-07-01-preview/generated.json
index 6e985fdceff9..86c41bc821ef 100644
--- a/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-07-01-preview/generated.json
+++ b/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-07-01-preview/generated.json
@@ -474,7 +474,7 @@
         },
         "content_filter_results": {
           "$ref": "#/definitions/ContentFilterResults",
-          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it \nhas been detected, as well as the severity level (very_low, low, medium, high-scale that \ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
+          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it\nhas been detected, as well as the severity level (very_low, low, medium, high-scale that\ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
           "x-ms-client-name": "contentFilterResults"
         }
       },
@@ -506,7 +506,7 @@
         },
         "prompt_filter_results": {
           "type": "array",
-          "description": "Content filtering results for zero or more prompts in the request. In a streaming request, \nresults for different prompts may arrive at different times or in different orders.",
+          "description": "Content filtering results for zero or more prompts in the request. In a streaming request,\nresults for different prompts may arrive at different times or in different orders.",
           "items": {
             "$ref": "#/definitions/PromptFilterResult"
           },
@@ -695,7 +695,7 @@
         },
         "content_filter_results": {
           "$ref": "#/definitions/ContentFilterResults",
-          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it \nhas been detected, as well as the severity level (very_low, low, medium, high-scale that \ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
+          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it\nhas been detected, as well as the severity level (very_low, low, medium, high-scale that\ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
           "x-ms-client-name": "contentFilterResults"
         },
         "logprobs": {
@@ -737,7 +737,7 @@
         },
         "prompt_filter_results": {
           "type": "array",
-          "description": "Content filtering results for zero or more prompts in the request. In a streaming request, \nresults for different prompts may arrive at different times or in different orders.",
+          "description": "Content filtering results for zero or more prompts in the request. In a streaming request,\nresults for different prompts may arrive at different times or in different orders.",
           "items": {
             "$ref": "#/definitions/PromptFilterResult"
           },
diff --git a/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-08-01-preview/generated.json b/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-08-01-preview/generated.json
index 6b4941e1c286..7ef06b6abcab 100644
--- a/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-08-01-preview/generated.json
+++ b/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-08-01-preview/generated.json
@@ -574,7 +574,7 @@
         },
         "content_filter_results": {
           "$ref": "#/definitions/ContentFilterResults",
-          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it \nhas been detected, as well as the severity level (very_low, low, medium, high-scale that \ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
+          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it\nhas been detected, as well as the severity level (very_low, low, medium, high-scale that\ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
           "x-ms-client-name": "contentFilterResults"
         }
       },
@@ -606,7 +606,7 @@
         },
         "prompt_filter_results": {
           "type": "array",
-          "description": "Content filtering results for zero or more prompts in the request. In a streaming request, \nresults for different prompts may arrive at different times or in different orders.",
+          "description": "Content filtering results for zero or more prompts in the request. In a streaming request,\nresults for different prompts may arrive at different times or in different orders.",
           "items": {
             "$ref": "#/definitions/PromptFilterResult"
           },
@@ -813,7 +813,7 @@
         },
         "content_filter_results": {
           "$ref": "#/definitions/ContentFilterResults",
-          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it \nhas been detected, as well as the severity level (very_low, low, medium, high-scale that \ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
+          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it\nhas been detected, as well as the severity level (very_low, low, medium, high-scale that\ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
           "x-ms-client-name": "contentFilterResults"
         },
         "logprobs": {
@@ -855,7 +855,7 @@
         },
         "prompt_filter_results": {
           "type": "array",
-          "description": "Content filtering results for zero or more prompts in the request. In a streaming request, \nresults for different prompts may arrive at different times or in different orders.",
+          "description": "Content filtering results for zero or more prompts in the request. In a streaming request,\nresults for different prompts may arrive at different times or in different orders.",
           "items": {
             "$ref": "#/definitions/PromptFilterResult"
           },
diff --git a/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-09-01-preview/examples/audio_transcription_object.json b/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-09-01-preview/examples/audio_transcription_object.json
index dc8c33a10a17..14fa046829a8 100644
--- a/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-09-01-preview/examples/audio_transcription_object.json
+++ b/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-09-01-preview/examples/audio_transcription_object.json
@@ -4,7 +4,8 @@
   "parameters": {
     "endpoint": "{endpoint}",
     "api-version": "2023-09-01-preview",
-    "deploymentId": "<deployment-id>"
+    "deploymentId": "<deployment-id>",
+    "body": "---multipart-boundary\nContent-Disposition: form-data; name=\"file\"; filename=\"file.wav\"\nContent-Type: application/octet-stream\n\nRIFF..audio.data.omitted\n---multipart-boundary--"
   },
   "responses": {
     "200": {
diff --git a/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-09-01-preview/generated.json b/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-09-01-preview/generated.json
index 12fdd77df2d2..121552c858f9 100644
--- a/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-09-01-preview/generated.json
+++ b/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-09-01-preview/generated.json
@@ -591,7 +591,7 @@
         "file": {
           "type": "string",
           "format": "byte",
-          "description": "The audio data to transcribe. This must be the binary content of a file in one of the supported media formats:\nflac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm."
+          "description": "The audio data to transcribe. This must be the binary content of a file in one of the supported media formats:\n flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm."
         },
         "response_format": {
           "$ref": "#/definitions/AudioTranscriptionFormat",
@@ -775,7 +775,7 @@
         "file": {
           "type": "string",
           "format": "byte",
-          "description": "The audio data to translate. This must be the binary content of a file in one of the supported media formats:\nflac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm."
+          "description": "The audio data to translate. This must be the binary content of a file in one of the supported media formats:\n flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm."
         },
         "response_format": {
           "$ref": "#/definitions/AudioTranslationFormat",
@@ -1087,7 +1087,7 @@
         },
         "content_filter_results": {
           "$ref": "#/definitions/ContentFilterResults",
-          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it \nhas been detected, as well as the severity level (very_low, low, medium, high-scale that \ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
+          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it\nhas been detected, as well as the severity level (very_low, low, medium, high-scale that\ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
           "x-ms-client-name": "contentFilterResults"
         }
       },
@@ -1119,7 +1119,7 @@
         },
         "prompt_filter_results": {
           "type": "array",
-          "description": "Content filtering results for zero or more prompts in the request. In a streaming request, \nresults for different prompts may arrive at different times or in different orders.",
+          "description": "Content filtering results for zero or more prompts in the request. In a streaming request,\nresults for different prompts may arrive at different times or in different orders.",
           "items": {
             "$ref": "#/definitions/PromptFilterResult"
           },
@@ -1326,7 +1326,7 @@
         },
         "content_filter_results": {
           "$ref": "#/definitions/ContentFilterResults",
-          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it \nhas been detected, as well as the severity level (very_low, low, medium, high-scale that \ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
+          "description": "Information about the content filtering category (hate, sexual, violence, self_harm), if it\nhas been detected, as well as the severity level (very_low, low, medium, high-scale that\ndetermines the intensity and risk level of harmful content) and if it has been filtered or not.",
           "x-ms-client-name": "contentFilterResults"
         },
         "logprobs": {
@@ -1368,7 +1368,7 @@
         },
         "prompt_filter_results": {
           "type": "array",
-          "description": "Content filtering results for zero or more prompts in the request. In a streaming request, \nresults for different prompts may arrive at different times or in different orders.",
+          "description": "Content filtering results for zero or more prompts in the request. In a streaming request,\nresults for different prompts may arrive at different times or in different orders.",
           "items": {
             "$ref": "#/definitions/PromptFilterResult"
           },
@@ -1945,7 +1945,7 @@
     "AudioTranscriptionOptions.file": {
       "name": "file",
       "in": "formData",
-      "description": "The audio data to transcribe. This must be the binary content of a file in one of the supported media formats:\nflac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm.",
+      "description": "The audio data to transcribe. This must be the binary content of a file in one of the supported media formats:\n flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm.",
       "required": true,
       "type": "string",
       "format": "byte",
@@ -2034,7 +2034,7 @@
     "AudioTranslationOptions.file": {
       "name": "file",
       "in": "formData",
-      "description": "The audio data to translate. This must be the binary content of a file in one of the supported media formats:\nflac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm.",
+      "description": "The audio data to translate. This must be the binary content of a file in one of the supported media formats:\n flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm.",
       "required": true,
       "type": "string",
       "format": "byte",