Bring your own key
Don’t want to run a model on your device, or want a top-end model? Bring your own key. Same IDE, same careful approach — OIOXO calls your provider directly from your browser. Your key is stored in your browser only and is never sent to an OIOXO server.
Supported providers
| Provider | Example models |
|---|---|
| OpenAI | gpt-4.1, gpt-4.1-mini, gpt-4o, gpt-4o-mini, o3, o3-mini, o4-mini |
| Anthropic (Claude) | claude-opus-4-8, claude-sonnet-4-6, claude-haiku-4-5, claude-fable-5 |
| Google (Gemini) | gemini-2.0-flash, gemini-2.5-pro, gemini-2.5-flash |
| OpenRouter | any model on OpenRouter (live list) |
| Groq | llama-3.3-70b-versatile, qwen-2.5-coder-32b |
| DeepSeek | deepseek-chat, deepseek-reasoner |
| Mistral | mistral-large-latest, codestral-latest |
| xAI (Grok) | grok-2-latest, grok-beta |
| Together AI | Qwen 2.5 Coder, Llama 3.3, DeepSeek V3 |
| Fireworks AI | Qwen 2.5 Coder, Llama 3.3, DeepSeek V3 |
| Cerebras | llama-3.3-70b, llama3.1-8b (fastest inference) |
| Ollama (local) | any model you've pulled — runs on your own machine, no key |
| Custom | any OpenAI-compatible endpoint (your base URL + model) |
Setting it up
- Choose “Use your own API key — no download” when picking a model (or switch any time from the model settings).
- Paste your key — OIOXO recognizes the provider from the key format automatically.
- Pick a model from the curated list (or the provider’s live list where available).
- OIOXO makes one tiny test call to confirm the key works, then you’re building.
Your bill, up to 90% smaller
OIOXO sends each request in a way your provider can reuse, so you pay for far fewer words. The meter shows words spent and words saved — including the dollar value. Free accounts get a monthly savings allowance; Pro removes the cap.
You decide how hard it tries: a Token saver dial in Settings goes from Off (send everything as-is) to Balanced (the default — real savings, full quality) to Aggressive (the leanest prompts and replies when you're watching the bill closely).
Ollama
Already running Ollama on your machine? OIOXO finds it at localhost:11434 and can use any model you've downloaded — strong local coding with no key and nothing sent to the cloud.
Your requests go from your browser straight to your provider. OIOXO never passes them through its own servers, never sees your key, and never sees your code along the way.