SN1 - APEX

Getting started - LLM Prompting on Macrocosmos

Welcome to APEX , the first live Large Language Model application on the Bittensor network and the heart of the Macrocosmos Constellation platform.

What is APEX?

APEX is a specialized LLM app focused on high-quality language model inference. Powered by validators and a constantly evolving pool of miners, it allows users:

  • Generate LLM completions using top open-source models

  • Retrieve real-time web results

  • Leverage advanced inference modes like multi-step reasoning or test-time thinking

Try the visual interface at Mission Command — a control center for LLM testing and prompting.

Base URL: https://sn1.api.macrocosmos.ai

The SN1 APEX API serves as access point for interacting with Macrocosmos's advanced language model inference capabilities. It performs chat completions and decentralized web searches through its primary endpoints: POST /v1/chat/completions and POST /web_retrieval. This API leverages a decentralized network of miners and validators to process requests, ensuring high-quality and reliable outputs. Developers can utilize the Macrocosmos SDK, specifically the ApexClient, to integrate these capabilities into their applications.

How the APEX API Works

At its core, the APEX API has two main endpoints:

  • POST /v1/chat/completions – Send chat-style messages to receive a generated response

  • POST /web_retrieval – Perform decentralized web search via multiple miners

Each request is routed to miners across the subnet based on task requirements, model filters, and optional parameters you provide. Validators coordinate execution, verify quality, and return reliable completions.

Models Available

SN1 currently supports the following open-source models:

  • Meta-Llama-3.1-70B-Instruct-AWQ-INT4

  • Mistral-Small-3.1-24B-Instruct

...with Google's Gemma 3 27B on the roadmap.

For more details, see our Supported Modelspage.

Rate limits

Regular API keys: Limited to 100 requests per hour Validator key: Limited to 1000 requests per hour

Need Help?

If you have any questions or require support, please message us in our official Macrocosmos Discord server . You can drop a message in the #apex-sn1 channel.

Last updated