Skip to content

LLM Provider Gateway vs Plumrocket AI Connector: one AI layer for a Magento store

Both put a store's AI keys and models in one place for other modules to use. Plumrocket adds a REST API, Hugging Face and Magento 2.3; LLM Provider Gateway is free and adds per-call cost, daily spend caps per module and local models.

Comparison

Every AI extension for Magento tends to arrive with its own API key field, its own model setting and no view of what it spends. A shared AI layer fixes that: keys and models are configured once, and other modules call the layer instead of a vendor. Plumrocket AI Connector and LLM Provider Gateway, which is ours, are both built for that job.

Everything about Plumrocket below was read on 2026-10-08 from its product page, its configuration documentation and its Adobe Commerce Marketplace listing. Where we say its documentation does not describe something, that is what we read, not a test of the product. Prices change; check the vendor's page before you buy.

The short answer

Choose Plumrocket AI Connector in any of these cases:

  • systems outside Magento must call the models through a REST API;
  • you want Hugging Face models;
  • you run Magento 2.3;
  • you already use Plumrocket modules that build on it.

Choose LLM Provider Gateway in any of these cases:

  • you want to know what each call cost and to cap each module's daily spend;
  • you want a different model for each module that uses AI;
  • you want to run a local model through Ollama;
  • you want to start at no cost.

Side by side

LLM Provider Gateway Plumrocket AI Connector
Price and licence Free of charge, with all future updates; commercial licence, not open source USD 99 one-time on plumrocket.com, with a year of updates and support. On the Adobe Commerce Marketplace an annual subscription: USD 99 the first year, then 49 (Open Source); 199, then 99 (Adobe Commerce)
Providers Anthropic, OpenAI, Google Gemini, OpenRouter, Ollama, Stability AI, Voyage AI, TypeSafe Claude, ChatGPT, Gemini, OpenRouter, Hugging Face
Local models Yes, through Ollama, keyless Not described in its documentation
Model choice Per consuming module, and per capability where the module offers it; optional fallback chain for chat Its documentation describes one default model per provider
Cost tracking Tokens, latency, status and estimated USD cost for every call, in a Usage Dashboard Its log documentation lists integration, model, status, prompt, response and time, without tokens or cost
Spend limits Daily USD cap per module plus a global default, and a master switch A maximum-tokens setting per request
Unsupported features A request that needs structured output, tools, images or streaming on a model that lacks it is refused before any spend Not described
Access for code PHP contracts for Magento modules; no REST, GraphQL or batch API A REST API and a PHP service interface
Magento versions 2.4.7 - 2.4.9, PHP 8.1 - 8.5 2.3 - 2.4.9
What uses it today Our AI Field Validator and FAQ Kit AI Generator Not stated on the pages we read
Latest version 1.2.0 1.0.3, 2026-09-27
Reviews New, none yet 5.0 from 1 on plumrocket.com

Where Plumrocket is the better choice

It reaches further outside Magento. Its REST API lets code outside Magento send requests through the same configuration, where LLM Provider Gateway offers PHP contracts to Magento modules only. It supports Hugging Face models and Magento 2.3, which we do not. Its one-time price on its own site includes a year of updates and support.

Where LLM Provider Gateway differs

It counts what it spends. Every call is logged with its tokens and an estimated cost in US dollars, computed from a price table you can correct. Each consuming module can be given a daily spend cap, and one switch stops every request at the gateway. The cap is a soft ceiling: the call that crosses it completes and the next one is refused, and our FAQ says so. Plumrocket's documentation describes a maximum-tokens setting per request, which limits the size of one answer rather than a module's total spend.

Each module gets its own model. A cheap model for one task and a strong one for another, changed in the admin with no code change in the module.

It can run without a cloud provider. Ollama is supported as a local, keyless provider, and the same module code runs against it.

It is free. A gateway only pays for itself through the modules built on it, so it costs nothing. It is free of charge but not open source: the source ships readable in the package, under the same commercial licence as our other extensions.

Amasty, for reference

Amasty's AI modules share an "AI Core" configuration with three options: Amasty's own AI service, available with an active product subscription, your own OpenAI key, or your own Gemini key. It serves Amasty's modules and is sold as part of them, for example the ChatGPT AI Content Generator at USD 189 on Community. It is not a stand-alone layer for other vendors' modules.

Sources, read 2026-10-08

How LLM Provider Gateway works is in its documentation; the extension is on its product page.