DEV.co
Open-Source DevOps · Capsize-Games

airunner

AI Runner is a desktop application for offline AI art generation and conversational AI companions, built with privacy in mind so everything runs locally without internet or API keys. It combines image generation (SDXL, Z-Image Turbo), voice conversation (TTS/STT), memory persistence, and a layered canvas interface.

Source: GitHub — github.com/Capsize-Games/airunner
1.3k
GitHub stars
99
Forks
Python
Primary language
GPL-3.0
License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

FieldValue
RepositoryCapsize-Games/airunner
OwnerCapsize-Games
Primary languagePython
LicenseGPL-3.0 — OSI-approved
Stars1.3k
Forks99
Open issues5
Latest releasev5.6.1 (2025-12-11)
Last updated2026-07-08
Sourcehttps://github.com/Capsize-Games/airunner

What airunner is

Python-based offline inference engine using llama.cpp for LLM inference, stable-diffusion derivatives for image generation, and voice synthesis/recognition via local models. Desktop UI built with PySide6, with optional headless HTTP API and Docker deployment support. GPU-accelerated via CUDA for NVIDIA hardware.

Quickstart

Get the airunner source

Clone the repository and explore it locally.

terminalbash
git clone https://github.com/Capsize-Games/airunner.gitcd airunner# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Privacy-First Creative Workflows

Organizations or individuals handling sensitive creative content (art, design, concepts) requiring zero data transmission outside their infrastructure.

Offline AI Companion Deployments

Enterprise or consumer environments where persistent, personalized conversational AI with memory and voice interaction must operate without cloud dependency.

Edge AI for Resource-Constrained Environments

Deployments where network latency, bandwidth limitations, or air-gapped systems require fully self-contained inference—studio environments, field work, offline research.

Implementation considerations

  • GPU memory and inference latency scale inversely with model size; RTX 3060 baseline may require quantized (Q8) models; RTX 5080+ recommended for SDXL in acceptable time.
  • Installation requires system-level CUDA toolkit, MeCab (for Japanese TTS), and multiple language-pack dependencies; Docker Compose or pre-built bundle installers strongly recommended over manual setup.
  • Database schema migrations via Alembic are manual if upgrading local installs; no automatic schema versioning or rollback strategy documented.
  • Voice STT only confirmed for English; multi-language TTS present but STT missing for Spanish, French, Chinese, Korean despite language-aware GUI.
  • Model management via HuggingFace and Civitai is manual (no auto-update); requires careful disk space management for large SDXL and LLM artifacts.

When to avoid it — and what to weigh

  • Need Multi-User Collaboration — AI Runner is single-machine focused; no built-in multi-user sync, shared state, or team workspace features.
  • Require Minimal Hardware Footprint — Minimum 16 GB RAM, 22–100 GB+ storage, and dedicated GPU (RTX 3060+) make this unsuitable for embedded, mobile, or lightweight edge devices.
  • Expect Production SLA/Support — Community-driven open-source project with no commercial support contract or guaranteed uptime guarantees; latest release is 7+ months old.
  • GPL-Incompatible Derivative Products — GPL-3.0 license requires any derivative work to be GPL-licensed; proprietary or other-license integration is legally complex—requires legal review.

License & commercial use

GPL-3.0 (GNU General Public License v3.0). This is a copyleft license requiring any derivative work, modification, or linked executable to be distributed under GPL-3.0 with source code available. Proprietary or closed-source integrations are not permitted without legal exemption.

GPL-3.0 permits commercial use of the unmodified software itself, but any commercial product or derivative that incorporates or links AI Runner must be GPL-3.0 licensed and open-source. This severely restricts commercial SaaS, proprietary plugins, or closed-source wrappers. Internal business use of the unmodified application is permitted. Strong legal review is required before any commercial deployment or integration.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

SignalAssessment
MaintenanceActive
DocumentationAdequate
License clarityClear
Deployment complexityHigh
DEV.co fitPossible
Assessment confidenceHigh
Security considerations

Project includes configurable NSFW filters and prompt classifiers for illegal content; no third-party security audit documented. All inference and data remain local, eliminating cloud-side injection risks. Cryptography for local conversation storage is not explicitly mentioned. Dependency pinning and supply-chain security practices (SBOM, signature verification) are not evident. Review dependency update cadence and vulnerability disclosure process before production use.

Alternatives to consider

LM Studio / Ollama

Lighter-weight, modular LLM-only inference engines with simpler setup and smaller resource footprint; lack integrated voice/art and desktop UI.

ComfyUI / Stable Diffusion Web UI

Mature node-graph or web-based image generation workflows with larger community and plugin ecosystem; require separate LLM and voice tooling.

LocalAI

Minimal self-hosted LLM/image API server supporting OpenAI-compatible endpoints; no voice, companion, or canvas UI—better for headless backend use.

Software development agency

Build on airunner with DEV.co software developers

AI Runner enables offline image generation, voice-driven chat companions, and creative workflows with zero cloud dependency. Evaluate GPU requirements, GPL-3.0 licensing constraints, and operational complexity before piloting. Contact Devco for architecture review and deployment planning.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

Related on DEV.co

Explore the category and the services that help you build with it.

airunner FAQ

Can I use AI Runner in a commercial product?
Only if your product and all its source code are GPL-3.0 licensed and open-source. Proprietary wrappers, SaaS, or closed-source derivatives are not permitted without GPL exemption (unlikely). Internal business use of the unmodified app is allowed. Consult legal counsel before commercial deployment.
What hardware is realistically required?
Minimum RTX 3060, 16 GB RAM, and 22–100 GB storage. Recommended: RTX 5080, 32 GB RAM, SSD. Quantized models (Q8) on RTX 3060 may produce 30–60s latency per generation; RTX 5080 cuts this to 5–10s. CPU-only inference is not practical for real-time use.
Is multi-language STT supported?
English STT only (documented as complete). TTS covers English, Japanese, Spanish, French, Chinese, Korean, but STT for non-English languages is not available. LLM supports all listed languages.
How do I run this in production / at scale?
Single-instance Docker Compose works for dev/small deployments. For production: pin exact image versions, manage persistent volumes for models/db, implement external health monitoring, and handle GPU/resource scaling externally (e.g. Kubernetes with GPU scheduling). Headless API is basic; consider API gateway and load-balancing outside AI Runner. No distributed inference or worker queue framework is built in.

Software development & web development with DEV.co

DEV.co is a software development agency delivering custom software development services to companies building on open source. Our software developers and web developers design, integrate, and ship production systems — spanning web development, APIs, AI, data, and cloud. If airunner is part of your open-source devops roadmap, our team can implement, customize, migrate, and maintain it.

Ready to Deploy Private AI Locally?

AI Runner enables offline image generation, voice-driven chat companions, and creative workflows with zero cloud dependency. Evaluate GPU requirements, GPL-3.0 licensing constraints, and operational complexity before piloting. Contact Devco for architecture review and deployment planning.