separate snippets

docs/api-inference/tasks/chat-completion.md

@@ -79,6 +79,7 @@ curl 'https://api-inference.huggingface.co/models/google/gemma-2-2b-it/v1/chat/c
 </curl>
 
 <python>
+With huggingface_hub client:
 ```py
 from huggingface_hub import InferenceClient
 
@@ -102,6 +103,7 @@ for chunk in stream:
     print(chunk.choices[0].delta.content, end="")
 ```
 
+With openai client:
 ```py
 from openai import OpenAI
 
@@ -132,6 +134,7 @@ To use the Python client, see `huggingface_hub`'s [package reference](https://hu
 </python>
 
 <js>
+With huggingface_hub client:
 ```js
 import { HfInference } from "@huggingface/inference"
 
@@ -159,6 +162,7 @@ for await (const chunk of stream) {
 }
 ```
 
+With openai client:
 ```js
 import { OpenAI } from "openai"
 
@@ -233,6 +237,7 @@ curl 'https://api-inference.huggingface.co/models/meta-llama/Llama-3.2-11B-Visio
 </curl>
 
 <python>
+With huggingface_hub client:
 ```py
 from huggingface_hub import InferenceClient
 
@@ -267,6 +272,7 @@ for chunk in stream:
     print(chunk.choices[0].delta.content, end="")
 ```
 
+With openai client:
 ```py
 from openai import OpenAI
 
@@ -308,6 +314,7 @@ To use the Python client, see `huggingface_hub`'s [package reference](https://hu
 </python>
 
 <js>
+With huggingface_hub client:
 ```js
 import { HfInference } from "@huggingface/inference"
 
@@ -346,6 +353,7 @@ for await (const chunk of stream) {
 }
 ```
 
+With openai client:
 ```js
 import { OpenAI } from "openai"
 

docs/api-inference/tasks/image-text-to-text.md

@@ -45,6 +45,7 @@ curl https://api-inference.huggingface.co/models/meta-llama/Llama-3.2-11B-Vision
 </curl>
 
 <python>
+With huggingface_hub client:
 ```py
 import requests
 
@@ -68,6 +69,7 @@ for chunk in stream:
     print(chunk.choices[0].delta.content, end="")
 ```
 
+With openai client:
 ```py
 import requests
 

scripts/api-inference/scripts/generate.ts

@@ -100,22 +100,22 @@ const TASKS_DATA = (await response.json()) as any;
 ///////////////////////
 
 const formatSnippets = (result: snippets.types.InferenceSnippet | snippets.types.InferenceSnippet[], defaultClient: string, language: string): string => {
-  // For single snippet, return just the content
+  // For single snippet, just wrap with code block
   if (!Array.isArray(result) || result.length === 1) {
     const snippet = Array.isArray(result) ? result[0] : result;
     return `\`\`\`${language}\n${snippet.content}\n\`\`\``;
   }
 
-  // For multiple snippets, wrap each one in its own code block
+  // For multiple snippets, add description and wrap each one
   return result
-    .map(snippet => 
-      `\`\`\`${language}\n${snippet.content}\n\`\`\``
-    )
+    .map(snippet => {
+      const client = snippet.client || defaultClient;
+      return `With ${client} client:\n\`\`\`${language}\n${snippet.content}\n\`\`\``;
+    })
     .join('\n\n');
 };
 
 
-
 const GET_SNIPPET_FN = {
   curl: (modelData: any, token: string) => {
     const result = snippets.curl.getCurlInferenceSnippet(modelData, token);