docs: Add AI/ML API example

[D1m7asis]

🔄 Summary

This PR adds support for AI/ML API, an OpenAI-compatible provider offering 300+ models (e.g., DeepSeek, Gemini, ChatGPT), to the E2B Code Interpreter examples.

---

✅ What's Included

• 🧪 examples/aimlapi-js/: TypeScript example using E2B sandbox + AI/ML API via OpenAI SDK
• 🧪 examples/aimlapi-python/: Python version using openai and e2b-code-interpreter
• 📊 Uses a CSV dataset to generate and run Python code (e.g., linear regression)
• 📄 Added .env.template, README.md, and dependency setup for each variant
• 🔗 Linked AI/ML API to the providers table in root README.md with proper UTM tracking

---

Let me know if you'd like the examples styled/formatted differently — happy to iterate!

[mishushakov]

Thanks. I will be reviewing this over the weekend.

[D1m7asis]

@mishushakov hi! Can i get a review here pls? Thanks!

[mishushakov]

Hey @D1m7asis, some points:

1. I have tried running the JS example with my own API key and got this:

   ❌ Error: 2 | import { castToError } from "./core.mjs";
   3 | export class OpenAIError extends Error {
   4 | }
   5 | export class APIError extends OpenAIError {
   6 |     constructor(status, error, message, headers) {
   7 |         super(`${APIError.makeMessage(status, error, message)}`);
                             ^
   error: 403 status code (no body)

   What am I doing wrong?

2. Can you make sure the example is same for both JS and Python?

3. I see that the examples are using gpt-3.5-turbo - can you use a more recent model, as not a lot of people would be starting with GPT-3.5 today

[D1m7asis]

Hey @mishushakov, thanks for the feedback 🙌

I’ve fixed the issues:
✅ Handled null content from the API so examples don’t crash on filtered outputs.
✅ Updated both Python and TypeScript examples to use the same flow (two prompts: strawberry count + linear regression with data.csv).
✅ Switched to a recent model (openai/gpt-5-chat-latest) instead of gpt-3.5-turbo.
✅ Verified both examples run fine with the same logic and produce consistent results.

Everything works now and the examples are identical across both languages 🚀

[OctavianTheI]

@mishushakov @tizkovatereza hi guys! a reminder to check on the PR :)

[mishushakov]

left some comments regarding getting the stdout of the sandbox

[D1m7asis]

@mishushakov thanks for the review 🙌
Before making the changes, I’d like to check, does e2b have a Slack for integration discussions?
If yes, could you please share an invite/connect to sergey@aimlapi.com (Sergey Nuzhny, DevRel + CM at AIMLAPI).
We can follow up with more details there if needed.

[OctavianTheI]

@mishushakov let us know if there is any other changes that need to be implemented
Let's wrap this up within 1-2 weeks, and roll out the promo

[D1m7asis]

Hi! Following up on this cookbook example.

The PR is still mergeable but has stale requested changes around example consistency/model choice/output handling. We can refresh it with the current recommended AI/ML API model, keep the Python and JS examples aligned, and make the sandbox stdout handling match the project style.

Please let me know if you would still like this example in the cookbook and I’ll update the branch accordingly.