Community Labs

Hypura Ollama-Compatible Inference API

Hypura is a storage-tier-aware LLM inference scheduler for Apple Silicon that places model tensors across GPU, RAM, and NVMe tiers so models larger than physical memory can run. Running `hypura serve ` exposes a local Ollama-compatible HTTP API, making it a drop-in replacement for tooling that talks to Ollama. The server is local-first; there is no hosted endpoint.

GitHub

Documentation

📖
Documentation
https://github.com/t8/hypura#readme
📖
APIReference
https://github.com/t8/hypura#endpoints
📖
GettingStarted
https://github.com/t8/hypura#quick-start

Other Resources

🔗
SourceCode
https://github.com/t8/hypura
🔗
Conformance
https://raw.githubusercontent.com/api-evangelist/community-labs/refs/heads/main/conformance/community-labs-conformance.yml
🔗
APIsJSON
https://raw.githubusercontent.com/api-evangelist/community-labs/refs/heads/main/apis.yml

API entry from apis.yml

apis.yml Raw ↑ 
aid: community-labs:hypura
name: Hypura Ollama-Compatible Inference API
description: Hypura is a storage-tier-aware LLM inference scheduler for Apple Silicon that places model
  tensors across GPU, RAM, and NVMe tiers so models larger than physical memory can run. Running `hypura
  serve <model.gguf>` exposes a local Ollama-compatible HTTP API, making it a drop-in replacement for
  tooling that talks to Ollama. The server is local-first; there is no hosted endpoint.
humanURL: https://github.com/t8/hypura
baseURL: http://127.0.0.1:8080
tags:
- LLM Inference
- Local AI
- Ollama Compatible
- Apple Silicon
properties:
- type: SourceCode
  url: https://github.com/t8/hypura
- type: Documentation
  url: https://github.com/t8/hypura#readme
- type: APIReference
  url: https://github.com/t8/hypura#endpoints
- type: GettingStarted
  url: https://github.com/t8/hypura#quick-start
- type: Conformance
  url: conformance/community-labs-conformance.yml