將 EidosStack AI Gateway 的 API 工作遷入 Meldport

I retired EidosStack AI Gateway as a standalone product and moved its API work into Meldport.

Meldport brings model discovery and API access together. Its public catalog lets you compare published model IDs, context windows, capabilities, and token rates before choosing a route. A project key and balance cover the published catalog.

For language models, the gateway accepts the familiar Chat Completions request shape. Point an OpenAI-compatible client at https://api.meldport.com/v1, choose a published model ID, and send a request:

curl https://api.meldport.com/v1/chat/completions \
  -H "Authorization: Bearer $MELDPORT_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: request-123" \
  -d '{
    "model": "YOUR_MODEL_ID",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

The API key stays on your server. The docs cover key scope, request limits, retries, and billing.

Image, video, and audio generation use their own workflows. They can involve input uploads, asynchronous jobs, and result retrieval, so they need more than a chat completion endpoint.

Meldport is available to the public now with prepaid, usage-based pricing. The official launch is set for 13 November 2026.

If you try it, I would appreciate feedback on the quickstart, the model details you need to compare routes, and how media generation fits into your app.