Ollama (local models) with Empryo

Run open-weight models on your own machine and use them in Empryo. Ollama is built into Empryo and is detected at localhost:11434, with no API key and nothing sent to the cloud. With a local model, your whole session can run offline with no per-token cost.

What you get

Example models

qwen3-coder (30b, 480b), tool-calling, long-context coding
qwen3 (0.6b, 235b), general, with tools + thinking variants
gpt-oss (20b, 120b), OpenAI open-weight, reasoning + tools
devstral (24b), tuned for coding agents
deepseek-r1 (1.5b, 671b), open reasoning, tools + thinking
deepseek-v3 (671b), large MoE
llama3.3 (70b), tool-calling general model
gemma3 (270m, 27b), single-GPU, vision
mistral-small (22b, 24b), small-tier tool-calling
qwen2.5-coder (0.5b, 32b), lightweight code models

When to use it

Pick Ollama when privacy, working offline or zero per-token cost matter more than having the strongest model: regulated codebases, machines without internet access, or heavy experimenting where cloud bills would add up. The model runs on your hardware, and Empryo's Genome (its map of your repo) and its symbol editing also run locally, so nothing needs to leave your machine. The trade-off: local open-weight models are weaker than frontier cloud models at complex multi-step agentic work, and the larger, more capable options (qwen3-coder:480b, deepseek-v3:671b) need serious VRAM. For day-to-day local use, tool-calling models in the 24, 32B range (devstral:24b, qwen3-coder:30b, gpt-oss:20b) are the sweet spot, and Empryo's agent loop depends on solid tool-calling, so prefer models tagged for tools. If you want frontier quality, use a cloud provider for the ember (code-writing) slot and keep Ollama for cheaper research and summary slots via the task router.

Set it up

1. Install and run OllamaLink to this section

Install Ollama from ollama.com, then pull a model with strong tool-calling (Empryo's agent loop relies on it):

bash

ollama pull qwen3-coder:30b

Ollama serves an OpenAI-compatible API at http://localhost:11434 by default.

2. No API key neededLink to this section

Ollama is a built-in provider in Empryo. It is a local, auth-less server, so there is nothing to set with --set-key. Empryo auto-detects it at localhost:11434 and lists any models you have pulled.

3. Point Empryo at a remote or custom host (optional)Link to this section

If Ollama runs on another machine or port, set the OLLAMA_HOST environment variable before launching:

bash

OLLAMA_HOST=http://192.168.1.50:11434 empryo

4. Pick the modelLink to this section

Launch Empryo and press Ctrl+L to open the model picker, then choose your Ollama model. To verify detection from the shell:

bash

empryo --list-providers          # confirm Ollama shows up
empryo --list-models ollama      # list your pulled models

You can also pin it as the default in ~/.empryo/config.json:

json

{
  "defaultModel": "ollama/qwen3-coder:30b"
}

5. Mix local and cloud with the task router (optional)Link to this section

Empryo's per-slot task router lets you run a small local model for cheap, high-frequency jobs while keeping a stronger model (local or cloud) for code writing:

json

{
  "taskRouter": {
    "spark":   "ollama/qwen2.5-coder:7b",
    "semantic":"ollama/qwen2.5-coder:7b",
    "ember":   "ollama/qwen3-coder:30b"
  }
}

Keys or subscriptions

Ollama (local models) connects the way marked below. You can mix ways across task-router slots.

API keyPaste a key once. Empryo keeps it in your system keychain, and you pay the provider for what you use.
Sign inApprove Empryo in your browser and use a plan or credits you already have. There is no key to copy.
On your machineOllama (local models)Run open models on your own computer. No key, no per-token bill, and nothing has to leave your machine.
Any compatible APIOne block of JSON config turns any OpenAI-compatible endpoint into a provider, including your own server.

Keep going

Download Empryo All providers