docs: Add AI/ML API example

.env.example

@@ -6,3 +6,4 @@ FIREWORKS_API_KEY=
 FIRECRAWL_API_KEY=
 GROQ_API_KEY=
 TOGETHER_API_KEY=
+AIML_API_KEY=

README.md

@@ -7,6 +7,8 @@ Example code and guides for building with [E2B SDK](https://github.com/e2b-dev/e
 
 Read more about E2B on the [E2B website](https://e2b.dev) and the official [E2B documentation](https://e2b.dev/docs).
 
+E2B works with any LLM that supports tool use. You can connect providers like OpenAI, Anthropic, Mistral, Groq, or custom OpenAI-compatible APIs such as the [AI/ML API](https://aimlapi.com/app/?utm_source=e2b&utm_medium=github&utm_campaign=integration).
+
 ## Examples
 
 **Hello World guide**
@@ -112,6 +114,13 @@ Read more about E2B on the [E2B website](https://e2b.dev) and the official [E2B
       <td><a href="https://github.com/e2b-dev/e2b-cookbook/tree/main/examples/watsonx-ai-code-interpreter-python">Python</a></td>
       <td><a href="https://github.com/e2b-dev/e2b-cookbook/tree/main/examples/watsonx-ai-code-interpreter-js">TypeScript</a></td>
     </tr>
+    <tr>
+      <td>AI/ML API</td>
+      <td> 300+ models</td>
+      <td>OpenAI-compatible API</td>
+      <td><a href="https://github.com/e2b-dev/e2b-cookbook/tree/main/examples/aimlapi-python">Python</a></td>
+      <td><a href="https://github.com/e2b-dev/e2b-cookbook/tree/main/examples/aimlapi-js">TypeScript</a></td>
+    </tr>
   </tbody>
 </table>
 

examples/aimlapi-js/.env.template

@@ -0,0 +1,3 @@
+# API keys for running the example
+AIML_API_KEY=your_aimlapi_api_key_here
+E2B_API_KEY=your_e2b_api_key_here

examples/aimlapi-js/.gitignore

@@ -0,0 +1,4 @@
+.env
+node_modules
+.eslintcache
+.eslintrc.json

examples/aimlapi-js/README.md

@@ -0,0 +1,67 @@
+# AI Code Execution with AI/ML API and E2B
+
+This example demonstrates how to run LLM-generated Python code using the [AI/ML API](https://aimlapi.com/app/?utm_source=e2b&utm_medium=github&utm_campaign=integration) and [E2B Code Interpreter SDK](https://e2b.dev).
+
+The AI performs a data analysis task on an uploaded CSV file, generates Python code with an AI/ML API model, and executes the code inside a sandboxed environment using E2B.
+
+---
+
+## 🔧 Setup
+
+### 1. Install dependencies
+
+```bash
+npm install
+````
+
+### 2. Setup environment
+
+Create `.env` file:
+
+```bash
+cp .env.template .env
+```
+
+Then set your:
+
+* `AIML_API_KEY`: Get it at [https://aimlapi.com/app/keys](https://aimlapi.com/app/keys/?utm_source=e2b&utm_medium=github&utm_campaign=integration)
+* `E2B_API_KEY`: Get it at [https://e2b.dev/docs/getting-started/api-key](https://e2b.dev/docs/getting-started/api-key)
+
+### 3. Run
+
+```bash
+npm run start
+```
+
+You will see:
+
+* The dataset gets uploaded
+* Prompt sent to the model
+* Code generated and executed in the cloud
+* A result saved as `image_1.png`
+
+![image](image_1.png)
+
+---
+
+## 🤖 Models
+
+This example defaults to **`openai/gpt-5-chat-latest`** via AI/ML API.
+You can switch to any OpenAI-compatible model available on AI/ML API (see the Models Directory in their docs).
+
+**Examples:**
+
+* `openai/gpt-5-chat-latest`
+* `openai/gpt-4o`
+* `google/gemini-1.5-flash`
+* `deepseek/deepseek-chat`
+* and many more
+
+> Ensure your API key has access to the chosen model.
+
+---
+
+## 🧠 Learn more
+
+* [AI/ML API Documentation](https://docs.aimlapi.com/?utm_source=e2b&utm_medium=github&utm_campaign=integration)
+* [E2B Code Interpreter Docs](https://e2b.dev/docs)

examples/aimlapi-js/aimlapi.ts

@@ -0,0 +1,197 @@
+import fs from 'node:fs'
+import path from 'node:path'
+import { Sandbox, Result, OutputMessage } from '@e2b/code-interpreter'
+import * as dotenv from 'dotenv'
+import OpenAI from 'openai'
+
+dotenv.config()
+
+const AIML_API_KEY = process.env.AIML_API_KEY || ''
+const E2B_API_KEY = process.env.E2B_API_KEY || '' // required by E2B SDK
+
+if (!AIML_API_KEY || !E2B_API_KEY) {
+  console.error('Missing API key(s). Please set AIML_API_KEY and E2B_API_KEY in your .env file.')
+  process.exit(1)
+}
+
+const openai = new OpenAI({
+  apiKey: AIML_API_KEY,
+  baseURL: 'https://api.aimlapi.com/v1',
+})
+
+const MODEL_ID = 'openai/gpt-5-chat-latest'
+
+// ---------- Prompts ----------
+const SYSTEM_STRAWBERRY = `
+You are a helpful assistant that can execute python code in a Jupyter notebook.
+Only respond with the code to be executed and nothing else.
+Respond with a Python code block in Markdown (\`\`\`python ... \`\`\`).
+`
+
+const PROMPT_STRAWBERRY = "Calculate how many r's are in the word 'strawberry'"
+
+const SYSTEM_LINEAR = `
+You're a Python data scientist. You are given tasks to complete and you run Python code to solve them.
+Information about the csv dataset:
+- It's in the \`/home/user/data.csv\` file
+- The CSV file uses "," as the delimiter
+- It contains statistical country-level data
+Rules:
+- ALWAYS FORMAT YOUR RESPONSE IN MARKDOWN
+- RESPOND ONLY WITH PYTHON CODE INSIDE \`\`\`python ... \`\`\` BLOCKS
+- You can use matplotlib/seaborn/pandas/numpy/etc.
+- Code is executed in a secure Jupyter-like environment with internet access and preinstalled packages
+`
+
+const PROMPT_LINEAR =
+  'Plot a linear regression of "GDP per capita (current US$)" vs "Life expectancy at birth, total (years)" from the dataset. Drop rows with missing values.'
+
+// ---------- Helpers ----------
+function extractPythonCode(markdown: string): string | null {
+  if (!markdown) return null
+
+  // 1) ```python ... ```
+  const rePython = /```python\s*([\s\S]*?)```/i
+  const m1 = rePython.exec(markdown)
+  if (m1 && m1[1]) return m1[1].trim()
+
+  // 2) ``` ... ```
+  const reAnyFence = /```\s*([\s\S]*?)```/
+  const m2 = reAnyFence.exec(markdown)
+  if (m2 && m2[1]) return m2[1].trim()
+
+  // 3) Fallback: treat whole string as code if it looks python-ish
+  if (/import |def |print\(|len\(/.test(markdown)) return markdown.trim()
+
+  return null
+}
+
+async function requestCode(systemPrompt: string, userPrompt: string): Promise<string> {
+  let response
+  try {
+    response = await openai.chat.completions.create(
+      {
+        model: MODEL_ID,
+        messages: [
+          { role: 'system', content: systemPrompt },
+          { role: 'user', content: userPrompt },
+        ],
+      },
+      {
+        headers: {
+          'HTTP-Referer': 'https://github.com/e2b-dev/e2b-cookbook',
+          'X-Title': 'e2b-cookbook:aimlapi-js',
+        },
+      }
+    )
+  } catch (err) {
+    throw new Error(`LLM request failed: ${String(err)}`)
+  }
+
+  const content = response?.choices?.[0]?.message?.content
+  if (content == null) {
+    throw new Error('Model returned null/empty content (possibly filtered). Try adjusting the prompt.')
+  }
+
+  const code = extractPythonCode(content)
+  if (!code) {
+    // Show what model returned for easier debugging
+    console.error('LLM response content:\n', content)
+    throw new Error('No Python code block found in LLM response.')
+  }
+  return code
+}
+
+async function runCodeInSandbox(code: string): Promise<{ results: Result[]; text: string; png?: Buffer }> {
+  const sandbox = await Sandbox.create()
+  try {
+    const exec = await sandbox.runCode(code)
+    if (exec.error) throw new Error(exec.error.value)
+
+    const results = exec.results ?? []
+    const first = (results[0] ?? {}) as any
+
+    const text = String(first?.text ?? '').trim()
+
+    let png: Buffer | undefined
+    if (first?.png) {
+      try {
+        png = Buffer.from(first.png, 'base64')
+      } catch {}
+    }
+
+    return { results, text, png }
+  } finally {
+    await sandbox.kill()
+  }
+}
+
+async function uploadDatasetIfExists(sandbox: Sandbox, localPath = './data.csv', targetName = 'data.csv') {
+  const p = path.resolve(localPath)
+  if (!fs.existsSync(p)) return false
+  const buf = fs.readFileSync(p)
+  await sandbox.files.write(targetName, buf)
+  return true
+}
+
+// ---------- Tests ----------
+export async function testStrawberry(): Promise<string> {
+  const code = await requestCode(SYSTEM_STRAWBERRY, PROMPT_STRAWBERRY)
+  const { text } = await runCodeInSandbox(code)
+  if (!text.includes('3')) {
+    throw new Error(`Expected '3' in output, got: ${JSON.stringify(text)}`)
+  }
+  return text
+}
+
+export async function testLinearRegression(imageOut = 'image_1.png'): Promise<string> {
+  const code = await requestCode(SYSTEM_LINEAR, PROMPT_LINEAR)
+
+  const sandbox = await Sandbox.create()
+  try {
+    const uploaded = await uploadDatasetIfExists(sandbox, './data.csv', 'data.csv')
+    if (!uploaded) {
+      console.warn('⚠️ data.csv not found next to aimlapi.ts — running anyway, code may fail if it expects the file.')
+    }
+
+    const exec = await sandbox.runCode(code)
+    if (exec.error) throw new Error(exec.error.value)
+
+    const first = (exec.results?.[0] ?? {}) as any
+    const text = String(first?.text ?? '').trim()
+
+    if (first?.png) {
+      try {
+        const png = Buffer.from(first.png, 'base64')
+        fs.writeFileSync(imageOut, png)
+        console.log(`✅ Image saved as ${imageOut}`)
+      } catch (e) {
+        console.warn('⚠️ Could not save image:', e)
+      }
+    } else {
+      console.warn('⚠️ No image result returned.')
+    }
+
+    return text
+  } finally {
+    await sandbox.kill()
+  }
+}
+
+// ---------- Entry ----------
+async function main() {
+  console.log('=== Strawberry test ===')
+  const s = await testStrawberry()
+  console.log(PROMPT_STRAWBERRY, '->', s)
+
+  console.log('\n=== Linear regression test ===')
+  const l = await testLinearRegression()
+  console.log('Linear regression output:\n', l)
+}
+
+if (require.main === module) {
+  main().catch((e) => {
+    console.error('❌ Error:', e)
+    process.exit(1)
+  })
+}