OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Infrastructure & Serving

Ferryai

Ferryai is an open-source PHP library that uses FFI to run ONNX, GGUF and RubixML model inference within a PHP process.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Ferryai — OpenViST project profile in AI Infrastructure & Serving.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Ferryai when a PHP application needs supported model inference in its own process and the team can meet the PHP and backend requirements.

Who it is for

  • PHP developers adding model inference to server-side applications.
  • Teams building PHP chat, embedding, classification or retrieval workflows.

Documented uses

  • Generate chat replies or stream tokens from a GGUF model.
  • Create text embeddings, store vectors and search them.
  • Run ONNX classification or moderation models.
  • Perform tabular prediction with CPU-native operations or RubixML models.
Before you choose

Requirements & limits

The repository is public and lists the MIT license. Ferryai is a library for developers to integrate, rather than a hosted inference service.

  • Requires PHP 8.3 or later.
  • ONNX and llama.cpp use native shared libraries through FFI; GPU operation has backend-specific dependencies.
  • The documented ONNX path can fall back to CPU when GPU providers are unavailable.
  • The repository describes the release as early and notes that the public API may change before 1.0.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up an ONNX embedding and vector-search flow

Configure Ferryai with an ONNX embedding model, then add and search vectors in its documented example.

Before you start

  • PHP 8.3 or later
  • Ferryai
  • An ONNX embedding model at a configured model path

Steps

  1. Configure the onnx backend and set the embedding model path.

  2. Call AI::embed() on text to obtain its vector.

  3. Create a vector store with AI::vector(), add the vector with an identifier and metadata, then search using an embedded query.

Expected outcome

A PHP code path that embeds text, stores a vector and retrieves matching vector-search results.

Guide limitations

  • The example does not document installation commands or how to obtain the model.
  • The example demonstrates the vector API but does not describe production data management.
Documentation 1 ↗
Source-based practical guide

Configure and call a GGUF chat model

Configure the llama backend with a GGUF model path and make a chat or streaming call.

Before you start

  • PHP 8.3 or later
  • A GGUF model file
  • The llama.cpp wrapper shared library, configured through FERRY_AI_LLAMA_WRAPPER or FERRY_AI_LLAMA_LIB and made available on PATH

Steps

  1. Set the backend to llama and configure its model_path to the GGUF file.

  2. Configure the wrapper location using one of the documented environment variables and add the relevant directory to PATH.

  3. Call AI::chat() for a complete reply or iterate over AI::stream() for tokens.

Expected outcome

PHP code configured to request chat output from a GGUF model, either as a full reply or a token stream.

Guide limitations

  • The README links to separate wrapper build instructions; this guide does not cover building the native wrapper.
  • The repository identifies the release as early and says its public API may change before 1.0.
Documentation 1 ↗
COMPARE OPTIONS

Compare options for Model inference and serving

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
FerryaiGenerate chat replies or stream tokens from a GGUF model.Self-hosted, ServerThe repository is public and lists the MIT license. Ferryai is a library for developers to integrate, rather than a hosted inference service.
Cloud AI SDKAdapt and run documented model recipes for computer-vision and NLP inference.Cloud, ServerThe GitHub repository is public. Its public visibility does not establish license terms.
TurboquantReplace a model’s existing inference cache with a compressed cache.Cloud, Linux, macOS, Self-hostedThe project describes itself as open source. The README documents installation through pip or from a source checkout.
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
MADEVAL/FerryAI · Official homepage title · retained in its source language
How we review sourcesSuggest a correction