EcoHash

Documentation in other languages: Chinese (Simplified)

EcoHash provides an OpenAI-compatible AI inference platform running on its own GPU cloud, with hosted LLM, text embedding, and rerank models that can be used directly from Dify.

Features

• Provides LLM (chat & vision), text-embedding, and rerank models in Dify.
• Includes predefined LLM models such as GLM-5.2, GLM-5-Turbo, DeepSeek-V4-Flash, Kimi-K2.6, MiniMax-M2.7, Qwen3-235B-A22B, qwen3-coder-30b-a3b-instruct, and qwen3-vl-8b-instruct.
• Includes predefined text embedding models such as jina-embeddings-v3, jina-embeddings-v4, and the qwen3-embedding series, plus the bge-reranker-v2-m3 reranker for RAG pipelines.
• Supports predefined model and customizable model configuration — add any model id from the EcoHash model catalog with just your API key.

Setup

1. Install this plugin from the Dify Marketplace.
2. Get your API key from the EcoHash console. Keys start with eco_.
3. In Dify, go to Settings → Model Provider → EcoHash and enter your API key.
4. Save the configuration.

The API key is the only credential required — the plugin connects to the EcoHash endpoint (https://api.ecohash.com/v1) automatically.

Usage

Select EcoHash as the model provider in Dify, choose an available model, and use it in applications, agents, or workflows.

For RAG, choose an EcoHash embedding model (e.g. jina-embeddings-v3) and the bge-reranker-v2-m3 reranker in your Knowledge Base settings.

Privacy

This plugin sends the inputs required by the selected operation to the EcoHash API. API keys are stored by Dify and never written to logs. See PRIVACY.md and the EcoHash privacy policy for details.

Source & support

• Source repository: https://github.com/ecohash-ai/ecohash-dify-plugin
• Issues and questions: GitHub Issues or the EcoHash docs
• Maintenance: the predefined model list is kept in sync with the live EcoHash catalog via scripts/sync_models.py (--check reports drift); catalog additions ship as patch releases of this plugin, and models not yet predefined can always be added immediately as customizable models.