Phengos

A native desktop AI workspace for people who want control over their AI workflow.

Use your own API keys with the models you already pay for, or run local models completely offline. Phengos gives you a structured workspace for prompts, files, projects, sessions, and AI-generated output — without a browser wrapper or subscription.

Bring your own provider. Run locally. Keep your work on your machine.

Why Phengos?

Use the models you choose

Phengos provides first-class integrations for Anthropic, OpenAI, and Mistral, with dedicated support maintained and tested by the Phengos team.

It also supports additional providers including OpenRouter, DeepSeek, Zhipu (GLM), Qwen, and Moonshot (Kimi).

Bring your own API key, choose your provider and model for each run, and pay the provider directly at its normal API rates. Phengos does not resell API access or add usage charges.

Run AI completely offline

Connect Phengos to Ollama or another local model server and keep your entire workflow on your machine.

• No API key.
• No cloud connection.
• No per-token charges.

The same workspace works for both cloud and local models.

Work with real files

AI work isn't always just text.

Attach reference and source files to a run as context. When a model produces a file, Phengos saves it directly to your project's output folder so you can use it outside the chat.

Keep every run

Your work is organized into projects and sessions, with each run saved to disk.

Review previous prompts and responses, see the model and token usage, check costs, reopen old work, export it, or use a previous run as the starting point for something new.

Your AI history isn't trapped in a chat interface.

Verify saved results

Phengos can seal saved runs and verify them later. This lets you determine whether a stored run is still exactly as it was when it was originally produced.

Build reusable workflows

Create reusable profiles and workflows for recurring tasks. Define the model's role, instructions, and conventions once instead of rebuilding the same context every time.

Native desktop software

Built as desktop software. Native C++ on our own rendering engine, drawn with Vulkan. No Electron, no embedded browser, no Node runtime, and no bundled package tree to keep patched.

---

Pro Edition

This itch.io listing is for Phengos Pro.

One-time purchase. Perpetual license for version 1.x. No subscription. No per-device charge.

Pro includes:

• Unlimited reference files per run
• Unlimited source files per run
• Persistent project and session history
• Multi-turn chat
• Run integrity checking
• Full cloud and local provider support

Community Edition

Phengos also has a free Community edition, licensed for commercial use.

Community supports the same AI providers as Pro, including both cloud and local models. Provider access is not restricted by edition. The differences are workflow limits.

Community includes:

• All cloud providers
• Local models
• 30 reference files per run
• 100 source files per run
• Commercial use

Pro adds:

• Unlimited reference files per run
• Unlimited source files per run
• Persistent project and session history
• Multi-turn chat
• Run integrity checking

The Community edition can be downloaded here or from GitHub: github.com/hyperoche/phengos

---

Requirements

Operating system

• Windows 10 or 11, 64-bit
• Linux, 64-bit
• Linux requires glibc 2.35 or newer
• Linux requires Wayland

Graphics

• A GPU with a working Vulkan driver

Phengos uses Vulkan for rendering and has no software rendering fallback. If your system does not have a working Vulkan driver, Phengos will not start.

Running in a virtual machine requires GPU passthrough

AI provider

• For cloud models, you need your own API key from a supported provider.
• For offline use, you need a local model server such as Ollama.

---

Before You Buy

Phengos is an AI client, not an AI model.

It does not contain its own model and does not provide cloud inference. You provide either:

• Your own API key for a supported provider, or
• A local model server running on your machine

Provider costs are separate from the Phengos purchase. Any API usage is billed directly by your chosen provider. Phengos does not include credits or resell API access.

Phengos currently works with text and code/file output. It does not generate images or produce binary office formats such as .xlsx or .docx.

Documentation is included with the application and covers installation, first run, configuration, and the features described above.

Your licence key is available from your purchase page after buying — click the request key button there.

---

Built for People Who Want Control

Phengos is for developers, researchers, technical users, and anyone who wants more than another browser-based AI chat window.

Use your own models. Use your own keys. Use your own files. Keep your runs. Or run everything locally.

Phengos puts the AI workflow on your desktop — and puts you in control.