Ctrlk
API KeyModelsPlaygroundGitHubGet Support
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Dify

Install the aimlapi.com plugin from the Dify Marketplace and use AI/ML API models across your Dify apps, workflows, and agents.

About

Dify is an open-source platform for building LLM applications — chat apps, agents, and visual workflows — with a built-in RAG pipeline and a model-provider layer that any app in the workspace shares.

AI/ML API ships as an official Dify Marketplace plugin, published as aimlapi/aimlapi. Install it from the Marketplace, paste one key, and every app in the workspace can use AI/ML API models.

Quick start

1

Open the Marketplace

In Dify, open Plugins and switch to the Marketplace tab.

2

Install aimlapi.com

Search for aimlapi.com and install the plugin.

3

Add your API key

Open Settings → Model Provider, find aimlapi.com, and paste your AI/ML API key.

4

Pick a model

Select a predefined model, or add a custom model by its exact AI/ML API model ID.

When to use AI/ML API with Dify

AI/ML API works well with Dify when you want:

  • one key and one provider entry instead of a separate credential per vendor

  • both the chat and embedding models a Dify RAG app needs from a single provider

  • to swap the model behind an app without touching the app's prompt or workflow

Prerequisites

Before you start, make sure you have:

Need a key first? Use API Key Management.

Supported model types

Dify model type
Used for

LLM

chat apps, agents, workflow LLM nodes

Text Embedding

indexing knowledge into Dify's RAG pipeline

Image, video, and audio models are not exposed through this plugin.

Model selection

The plugin ships a generated snapshot of the AI/ML API chat catalog — over 300 predefined models, hottest first — plus embedding models. Anything not in the list can be added as a custom model: choose Add Model, and enter the full AI/ML API model ID.

Good starting models

  • openai/gpt-5-5 — strong general use

  • anthropic/claude-sonnet-5 — long-form and agentic work

  • google/gemini-3.6-flash — lower latency and lower cost

  • openai/text-embedding-3-small — embeddings

For the full catalog, use All Model IDs.

Config checklist

Make sure these values are set:

  • plugin: aimlapi.com installed from the Dify Marketplace

  • API key: your AI/ML API key, saved in Settings → Model Provider

  • model: a predefined model, or a custom model with an exact AI/ML API model ID

Troubleshooting

I cannot find the plugin in the Marketplace

Search for aimlapi.com. On a self-hosted Dify, confirm that Marketplace access is enabled for the instance; otherwise install the plugin package manually from the plugin repository.

The key saves, but inference fails

A key with a zero balance can still be saved. If the balance is insufficient at inference time, the plugin returns a link to add credits — top up at aimlapi.com/app.

A model I want is not in the list

The predefined list is a snapshot of the catalog. Add the model as a custom model with its exact AI/ML API model ID, including the vendor prefix.

I get an auth error

Check that:

  • the key is valid at aimlapi.com/app/keys

  • the model ID matches the catalog exactly

  • the workspace uses the provider entry you configured, not a leftover custom OpenAI-compatible entry

Links

Previouscontinue.dev
NextElizaOS

Last updated

Was this helpful?