PHP-ORT: Machine learning inference for the web

> Every modern application needs intelligent features

I'm not so sure about that. And I don't even know how and for what I would use any of that.

> stay relevant in an AI-first world or risk obsolescence

Damn I'm doomed :)

There's also https://github.com/symfony/ai for a more LLM focused library which it seems https://github.com/php-llm/ migrated to.

Interesting idea! No matter what anyone says, PHP is still used today and will be used

Thank you for working on this! Since there was no option available, we were required to use Python with some API or just plain API for the major paid models. Now with ONNX we can load open-source model and use them.

Keep up the good work and I hope this will bring a lot more packages that will make this ecosystem a little more up-to-date!

That's super cool. But do young devs/engineers even touch PHP these days? I haven't coded in it seriously for over a decade. But keep it up!

This is great. Very rudimentary in this state but the right direction. What php also needs is a good lib for NLP in different languages

I wonder what Pieter Levels uses, since all of his apps are PHP based and he has a number of AI/ML offerings these days.

https://levels.io/

So basically NumPy, but for PHP. The benchmarks look good.

I don’t believe that 78% of websites run on PHP

Just what we needed....

If 78% of the web is PHP, then 1799% of the web must be javascript.

> Every modern application needs intelligent features

I'm not so sure about that. And I don't even know how and for what I would use any of that.

> stay relevant in an AI-first world or risk obsolescence

Damn I'm doomed :)

There's also https://github.com/symfony/ai for a more LLM focused library which it seems https://github.com/php-llm/ migrated to.

Check the photo in your pocket, are you fading? Quick, go write some AI before you disappear completely!

Keep up the good work and I hope this will bring a lot more packages that will make this ecosystem a little more up-to-date!

Just what we needed....

Interesting idea! No matter what anyone says, PHP is still used today and will be used

This is great. Very rudimentary in this state but the right direction. What php also needs is a good lib for NLP in different languages

I wonder what Pieter Levels uses, since all of his apps are PHP based and he has a number of AI/ML offerings these days.

https://levels.io/

He's mentioned (on twitter, and his Lex Fridman interview) relying on https://replicate.com/ and https://fal.ai/ as the workhorses for his AI stuff.

If 78% of the web is PHP, then 1799% of the web must be javascript.

So basically NumPy, but for PHP. The benchmarks look good.

Yeah, I was a little confused at all the talk about ML becoming a first class citizen in the language, when in reality it’s just a math library. (a seemingly good math library to be fair!)

I don’t know though, I feel like calling out to an API for doing ML stuff in PHP is always gonna make more sense. The compute power in services offered by AI companies are typically gonna be better than what you probably have on the server that’s running your PHP code. Unless your business is completely built around doing some kind of ML inference AND you went all in on PHP for the backend, then I guess it would make sense.

It’s still a cool extension. I myself still maintain multiple websites built with PHP. It’s not dead people!

I don’t believe that 78% of websites run on PHP

That's super cool. But do young devs/engineers even touch PHP these days? I haven't coded in it seriously for over a decade. But keep it up!

They're counting wordpress, drupal, various forums, etc. They make up a huge percentage of custom sites.

If you're talking only custom built stacks, the percentage is definitely lower, but probably still surprisingly high.

Its a "fact" PHP devs love. The truth is 95% of PHP is not actaully PHP, bit wordpress/drupal and the other cms's.

Its click, drag, click drag and the underlying tech has zero importance really.

78% of websites where it can be deduced from client side code what server side tech is used. W3tech published this statistic, and the whole community picked it up while ignoring that very important caveat.

Not a very meaningful statistic.

Anyone who cafés about security would stay away from PHP. Certainly more than 22% of the web.

I know several people who got burn out working in the js ecosystem and who went to php instead. Finally things are quiet, stable and fast. Older people I know don't really care (I worked with IBM software in the 80s/90s, talking about slow and bloated), but the younger ones get very upset with the churn, bloat, overhead, etc.

I don't know about trends among young developers, but I can tell you that the majority of the fintech world that isn't Microsoft centered (or ancient COBOL hieroglyphics) runs on PHP these days.

I ended up finding myself in finance because I had so much PHP experience from bespoke ecommerce and never left because it's everywhere.

Other language ecosystems have things I like better personally, but if you master PHP, you will never have a problem finding work.

about 6 months ago we hired a 24 aged guy as a php developer (with more than 3 year experience). so yes. with recent changes in PHP, like modernize stack, good frameworks (laravel, symfony, hyperf) and huge speed it gains from new runtimes like swoole, openswoole, frankenphp, it is really good stack to start new projects and have almost zero need for future migration to something else. i need to mention we have GO in our backend stack as well and no plan to migrate anything but in our case GO services can easily be replaced with new PHP runtimes and we would have no performance issues.

> But do young devs/engineers even touch PHP these days?

Of course! I'm 41 and still code in PHP. ;)

I know probably 10 or more young PHP devs and zero Rust devs (regardless of age).

PHP is great, especially with Laravel.

i know js devs switching to php, modern php is something quite different than the old one

PHP has been on a steady decline for the last 10-15 years. Its a slow process, but in the end it most likely will fade out.

He's mentioned (on twitter, and his Lex Fridman interview) relying on https://replicate.com/ and https://fal.ai/ as the workhorses for his AI stuff.

Not a very meaningful statistic.

They're counting wordpress, drupal, various forums, etc. They make up a huge percentage of custom sites.

If you're talking only custom built stacks, the percentage is definitely lower, but probably still surprisingly high.

Its a "fact" PHP devs love. The truth is 95% of PHP is not actaully PHP, bit wordpress/drupal and the other cms's.

Its click, drag, click drag and the underlying tech has zero importance really.

I don't know about trends among young developers, but I can tell you that the majority of the fintech world that isn't Microsoft centered (or ancient COBOL hieroglyphics) runs on PHP these days.

I ended up finding myself in finance because I had so much PHP experience from bespoke ecommerce and never left because it's everywhere.

Other language ecosystems have things I like better personally, but if you master PHP, you will never have a problem finding work.

> But do young devs/engineers even touch PHP these days?

Of course! I'm 41 and still code in PHP. ;)

Anyone who cafés about security would stay away from PHP. Certainly more than 22% of the web.

Cargo-culting bs that you heard from somewhere and never bothered to fact check? Shame on you.

Oh really? Why is that. Let's talk about globals from 2005.

I work in an innovation team in a tech company, we are free to choose our own stacks. I went for Laravel, one of the other engineering leads, RoR fan, went for NextJS. One of us is very happy with their batteries included stack, the other keeps telling me how much they miss RoR.

i know js devs switching to php, modern php is something quite different than the old one

I wish more people would. The eco system is so good.

PHP has been on a steady decline for the last 10-15 years. Its a slow process, but in the end it most likely will fade out.

PHP is great, especially with Laravel.

Cargo-culting bs that you heard from somewhere and never bothered to fact check? Shame on you.

I know probably 10 or more young PHP devs and zero Rust devs (regardless of age).

Rust seems like something everyone talks about but no one actually uses.

Oh really? Why is that. Let's talk about globals from 2005.

It’s not 2005? How is this even an argument? Age-old opinions of PHP are do not reflect the modern state of PHP and its ecosystem.

Yeah, I was a little confused at all the talk about ML becoming a first class citizen in the language, when in reality it’s just a math library. (a seemingly good math library to be fair!)

It’s still a cool extension. I myself still maintain multiple websites built with PHP. It’s not dead people!

I wish more people would. The eco system is so good.

It’s not 2005? How is this even an argument? Age-old opinions of PHP are do not reflect the modern state of PHP and its ecosystem.

Rust seems like something everyone talks about but no one actually uses.

The Inevitable Transformation

Software is changing faster than we've seen in 25 years. Machine learning isn't just becoming important, it's becoming essential. Every application, every website, every digital interaction will soon expect intelligent features as standard.

For millions of PHP developers who power the web, this creates an existential challenge: stay relevant in an AI-first world or risk obsolescence.

The Stakes: This isn't about technology preferences, it's about livelihoods. Mortgages, families, careers built on PHP expertise.

The Current Reality

PHP dominates the web: 78% of websites use PHP as their server-side language
ML is becoming essential: Every modern application needs intelligent features
The gap is growing: No first class machine learning support for PHP
The choices are terrible: Burden your stack with microservices, API calls, or other inefficient methodology (eg. FFI), or switch stacks.

=1 gflops) 1000×1000 float32 matrices #### CPU Acceleration 8x speedup with AVX2 #### GPU Acceleration ?? speedup with CUDA #### Hybrid Parallelism 100% CPU/GPU utilization ### Complementary, Not Competitive This is about finding PHP a place in an AI-first software engineering world, the one that's coming ... not only does AI need to be a first class citizen in the _minds of developers_, but also in the _languages they use_: * **Python:** Training, research, model development * **PHP:** Web inference, production serving, real-time processing * **ONNX:** The bridge between training and inference _This isn't about replacing Python for training!_ Technical Excellence -------------------- This isn't a proof of concept, it's production-ready infrastructure built to exacting standards. ### Core Architecture **PHP API Layer** - Clean, decoupled API: ORT\\Tensor, ORT\\Math functional namespace, ORT\\Model, ORT\\Runtime **Math Library** Frontend: Dispatch, scheduling, and scalar fallbacks (always available) Backend: Optimizations (CUDA, WASM, NEON, RISCV64, AVX512, AVX2, SSE4.1, SSE2) **ONNX Integration** ORT\\Model: Loading, metadata, and management ORT\\Runtime: Inference **ORT\\Tensor** - Immutable API; always available ### Key Technical Innovations * **[Immutable Tensors:](#)** Zero-copy, lock-free data sharing, efficient memory usage, and predictable performance * **[Dual Storage Class Tensors:](#)** ORT\\Tensor\\Persistent survive request cycles (and may be shared among threads, ie FrankenPHP), ORT\\Tensor\\Transient are request local * **[Memory Management:](#)** Aligned allocator and optimized memcpy for maximum performance potential * **[Thread Pool:](#)** Dedicated Slot scheduling implementation with alignment awareness * **[SIMD Optimization:](#)** Runtime detection with thread pinning ensures stability and predictability at scale * **[GPU Optimization:](#)** Optional GPU acceleration ensures saturation of all available parallel hardware * **[Type System:](#)** Schemas extracted from NumPy runtime; no guesswork, perfect compatibility * **[Type Promotion:](#)** Automatic conversion between types for seamless integration and maximum predictability * **[Zero Overhead Optimizations:](#)** Backend silently (opaque to the frontend) optimizes dispatch * **[Call Site Scaling:](#)** ORT\\Math\\scale provides a high degree of control over scaling at the call site * **[Modular Design:](#)** Math and ONNX systems are independent - use either or both as needed * **[Flexible Tensor Generation:](#)** ORT\\Tensor\\Generator provides flexible generation (lazy loading, random, etc) ### Performance Benchmarks ![Benchmark Chart](https://krakjoe.github.io/ort/bench/images/mul.png) All benchmarks are reproducible and available in the [bench](https://github.com/krakjoe/ort/tree/develop/dist/bench) directory. ### Getting Started wget https://.../v1.22.0/onnxruntime-linux-x64-1.22.0.tgz -O onnxruntime-linux-x64-1.22.0.tgz sudo tar -C /usr/local --strip-components=1 -xvzf /onnxruntime-linux-x64-1.22.0.tgz sudo ldconfig sudo apt-get install pkg-config phpize ./configure --enable-ort make sudo make install extension=ort.so get(); ### Real-World Impact * **Developer Retention:** Millions of PHP developers stay relevant * **Web Evolution:** Every PHP website becomes a potential ML application * **Accessibility:** ML inference becomes as common as database queries * **Innovation:** New applications emerge when barriers are removed ### Join the Evolution This isn't about revolution, it's about evolution. PHP has always adapted to stay relevant. From simple scripts to enterprise applications, and now to intelligent applications. **The Vision:** Every corner of the web becomes capable of intelligent behavior. PHP developers don't need to choose between staying with PHP and building the future, they can do both. Machine learning inference isn't a luxury anymore, it's becoming essential infrastructure. This is PHP's answer to the AI revolution.

Hacker Times

Hacker Times

PHP-ORT: Machine learning inference for the web

Discussion

Discussion

The Inevitable Transformation

The Current Reality