Explore/Products/idyl.inference

idyl.inference

by Idyl Labs

OpenAI-compatible inference for open-source models on the network.

01 Overview

idyl.inference is a hosted endpoint for open-source models running on the idyl.inference subnet. The models are already deployed; users get an API key and call the endpoint.

The surface follows the OpenAI API closely enough that existing clients can often be adapted by changing the base URL and API key.

Idyl handles model placement and provider routing underneath the endpoint, so teams consume one service rather than managing the machine layer directly.

02 Capabilities

Current shape.

Details published for this product.

01

OpenAI-compatible surface.

Uses familiar endpoints such as /v1/chat/completions, /v1/embeddings, and /v1/models.

02

Pre-deployed open models.

Curated catalog of open-source models — Llama, Mistral, Qwen, and more. No deployment, no model loading, no cold starts.

03

Works with existing clients.

SDKs and tools that speak the OpenAI protocol can often be adapted by changing the base URL and API key.

04

One API key.

Sign in, generate a key, start calling. No infrastructure setup. No project configuration. No region selection.

05

Distributed underneath.

Requests are routed across providers in the idyl.inference subnet automatically. Capacity scales with the network.

06

Managed model catalog.

Idyl runs and maintains the available models so teams can focus on the application layer.

03 Quickstart

Drop-in replacement.

Use the OpenAI SDK or any OpenAI-compatible client. Change the base URL, keep your existing code.

# Python — works with the OpenAI SDK as-is
from openai import OpenAI

client = OpenAI(
    base_url="https://api.inference.idyl.dev/openai/v1",
    api_key="your-api-key",
)

response = client.chat.completions.create(
    model="llama-3.1-70b-instruct",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)
04 Where it runs

Hosted on the idyl.inference subnet.

idyl.inference runs on idyl.inference. The subnet's policy and operator behavior are part of the environment behind the product.

06 Get listed

List a product.

If a product is built on Idyl and its current stage is explicit, it can be submitted for listing.

— Criteria

Listing criteria.

Products may be public, limited-access, or under development. The listing should make the current stage, access model, and relationship to Idyl explicit.

  • 01 Built on Idyl, or clearly being built on Idyl
  • 02 Current stage is explicit
  • 03 Access model is explicit
  • 04 Product page, docs, or contact path is real
[ Next ]

idyl.inference runs on idyl.inference.