jevlaya
English

Ollaya

ollaya-dev

A single Rust binary that pulls, serves and runs open decision models locally. A native /api/* for model management, plus a /v1/* endpoint implemented against the TypeSafe wire schema.

Platforms
linux · macos · windows
GPU
Not required
API compatibility
systemone-v1
License
Apache-2.0
Models it can run
16

Install

linux

curl -fsSL https://ollaya.dev/install.sh | sh

macos

curl -fsSL https://ollaya.dev/install.sh | sh

windows

irm https://ollaya.dev/install.ps1 | iex

127.0.0.1:11435

Documentation

Run open decision models on your own machine, behind a TypeSafe-compatible API

  • The CLI, the server and the model runners live in one binary — no Python environment to set up
  • /v1/* follows the TypeSafe wire schema, so existing SDK code switches by changing a base URL
  • Runs on CPU — the smaller models answer in a fraction of a second without a GPU
  • Speaks MCP, so an agent can make a decision before every routing or triage step
  • Compatibility is claimed by the project, pinned to the wire schema of typesafe-sdk 0.7.1 — not a TypeSafe endorsement
  • Young project: the repository was created 2026-09-23 and v0.9.0 shipped 2026-10-02, so commands and APIs can still move
  • Accuracy comes from whichever model you pull — the runtime itself decides nothing
  • The headline comparison with Jev comes from the project’s own material, not a controlled head-to-head

Data verified 2026-10-05