Aura
Kodine's curated model lineup, behind one API and one balance.
Kodine Aura is a hand-picked selection of coding models operated by the Kodine team.
Aura plugs into Kodine the same way any other provider does: sign in, create an API key, and start coding. Using it is entirely optional — bringing your own provider is always supported.
Background
Hundreds of models exist across dozens of providers, and every provider configures them differently. As a result, the very same model can behave in completely different ways depending on where you access it.
To address this, we took a few steps:
- Candidate models are benchmarked on real coding-agent workloads.
- Each configuration is validated before it goes live.
- Quality is monitored continuously after release, with better versions swapped in as they become available.
Kodine Aura is an AI gateway that exposes these models through a single API and a single balance.
How it works
Aura connects to Kodine the same way any other provider would.
- Sign in to Kodine Aura, enter your billing details, and copy an API key.
- In the TUI, run the
/connectcommand, choose Kodine Aura, and paste the key. - Run
/modelsin the TUI to browse the models we recommend.
Billing is per token, and you can top up your balance with as little as $5.
Endpoints
Our models are also reachable directly through these API endpoints.
The API base URL is https://api.kodine.net — the machine-facing host for the gateway (/v1/*, /aura/v1/*), auth (/auth/*), the CLI update feed, and the install scripts. kodine.net itself remains the human website.
| Model | Model ID | Endpoint | AI SDK Package |
|---|---|---|---|
| Kodine Max | kodine-max | https://api.kodine.net/v1/chat/completions | @ai-sdk/openai |
| Kodine Large | kodine-large | https://api.kodine.net/v1/chat/completions | @ai-sdk/openai |
| Kodine Code | kodine-code | https://api.kodine.net/v1/chat/completions | @ai-sdk/openai |
| Kodine Swift | kodine-swift | https://api.kodine.net/v1/chat/completions | @ai-sdk/openai |
| Kodine Small | kodine-small | https://api.kodine.net/v1/chat/completions | @ai-sdk/openai |
| Kodine Mini | kodine-mini | https://api.kodine.net/v1/chat/completions | @ai-sdk/openai |
Every endpoint speaks the OpenAI chat completions format. An Anthropic-compatible endpoint is available too, at https://api.kodine.net/v1/messages, so any agent that lets you set a custom base URL can talk to Aura.
In your Kodine config, the model id is written as kodine/<model-id>. So to use Kodine Large, you would set kodine/kodine-large.
Models
The full catalog of models and their metadata can be fetched from:
https://api.kodine.net/v1/modelsPricing
Aura is pay-as-you-go. The table below lists prices per 1M tokens.
| Model | Input | Output |
|---|---|---|
| Kodine Max | $6.00 | $30.00 |
| Kodine Large | $3.00 | $15.00 |
| Kodine Code | $2.00 | $10.00 |
| Kodine Swift | $1.00 | $4.00 |
| Kodine Small | $0.30 | $1.20 |
| Kodine Mini | $0.10 | $0.40 |
Kodine Small and Kodine Mini are part of the free tier, so you can try Kodine without topping up. A paid plan — or any balance top-up — unlocks the full lineup.
What you see is what you pay: no hidden multipliers, no processing fees.
Contact us with any questions.
Auto-reload
When your balance falls below $5, Aura tops it up automatically.
The auto-reload amount is configurable, and the feature can be switched off completely.
Monthly limits
Spending limits can be set per month for the whole workspace as well as for individual team members.
Say you cap monthly usage at $20 — Aura will stay within that $20 for the month. Keep in mind, though, that with auto-reload enabled, a balance dipping under $5 can still trigger charges beyond the $20 cap.
Privacy
Every Aura model runs on infrastructure with a strict zero-retention policy: your code is never stored, and never used for training.
For Teams
Aura works well for teams too: invite teammates, assign roles, curate which models the team can use, and more.
Roles
Teammates can be invited into your workspace with one of these roles:
- Admin: Manage models, members, API keys, and billing
- Member: Manage only their own API keys
To keep spending in check, admins can also assign monthly limits to each member.
Model access
Admins control which models are enabled for the workspace. Any request sent to a disabled model fails with an error.
Bring your own key
Your own API keys for other providers can be used side by side with Aura models.
Tokens spent through your own keys are billed by that provider directly — Aura doesn’t charge for them.
Web chat
Workspace members can chat with the workspace’s models right in the console, at /workspace/<id>/chat. It’s a ChatGPT-style interface — conversation sidebar, model picker, and streaming responses — handy for quick questions outside the terminal.
Goals
We built Kodine Aura to:
- Benchmark the strongest models for coding agents.
- Deliver consistent quality — no silent downgrades, no surprise reroutes.
- Keep pricing simple and public — a single page with token prices and nothing else.
- Avoid lock-in — Aura works with any agent, and Kodine works with any provider.