Gunamaya AI
Models & Providers
Gunamaya AI connects to the AI model you choose using your own API key — there's no bundled model, no markup, and no single point of failure if one provider has an outage.
Bring your own key
Instead of routing every request through a single company's servers, Gunamaya AI asks you to connect a model provider directly. This means:
You control cost
You're billed directly by your provider at their standard API rates — Gunamaya doesn't add a markup or subscription fee on top.
You control which model
Use a fast, cheap model for quick questions and a more capable one for hard problems, switching per conversation.
No single point of failure
If one provider is down or rate-limited, switch to another without waiting on anyone but yourself.
Your keys, your machine
API keys are stored using your operating system's secure credential store, never synced to a third-party server.
Connecting a provider
Open Model settings
From the chat panel, click the model selector above the input box, or open Gunamaya AI Settings → Model directly.
Choose a provider
The model catalog covers the major hosted API providers as well as OpenAI-compatible endpoints, so a self-hosted or lesser-known provider that speaks the same protocol works too.
Paste your API key
Keys are validated on save and stored securely — they're never written into your workspace files or included in anything sent back to a model as context.
Pick a default model
Set a default for new conversations, and override it per-conversation from the same selector whenever you want something different.
Different models for different jobs
A common setup is a fast, inexpensive model as your default for everyday questions and small edits, with a heavier reasoning-focused model kept one click away in the selector for the harder problems that benefit from it.
Local and self-hosted models
If your provider entry is an OpenAI-compatible HTTP endpoint — including locally-run inference servers — point Gunamaya AI at its base URL the same way you'd configure any other custom provider. This keeps every request on your own machine or network with no external API calls at all.
Switching providers mid-project
There's no lock-in step. Changing your default model, adding a second provider, or removing a key entirely are all done from the same Model settings page, and take effect on your very next message — no restart required.