AI for code
AI for code: chat, projects and an OpenAI-compatible API
Luger writes, reads and fixes code in the chat, builds whole projects with files and checks from a phone or a laptop, and serves the same models through an API that works with OpenAI clients.
In the chat
Paste a file or a snippet and get a review, tests, a refactor or an explanation. The model sees images: a screenshot of the error or a diagram is enough. Up to 1M tokens of context fits a large module whole.
A Think mode for hard reasoning and web search with sources for documentation and fresh APIs are one toggle away.
Projects from a phone
Describe the idea and the agent creates files, runs the steps, shows a preview and a task list. A one-page site, a script, a bot, a prototype, without opening a laptop. Download the result as an archive or continue in the next chat.
An OpenAI-compatible API
The /v1/chat/completions endpoint, streaming, tool calls, JSON replies. Change the base URL and the key, and Cursor, Continue or your own scripts work with Luger. Keys are issued in the account; each can carry a monthly cap.
Pricing is per token from the same balance as the chat: from $0.40 per 1M input tokens for luger-v1 and from $2 for luger-v2.
Without needless refusals
Security work such as CVE analysis, malware breakdowns and tooling for authorised tests gets done rather than lectured about. Describe the authorised environment and the goal; the only limit is the law.
Questions
Which models are available?
Luger 1 (fast and cheap), Luger 2 (the workhorse: code, files, tools), Luger 2 Pro (hard reasoning, up to 260k tokens per reply).
Does it work with Cursor?
Yes. The API follows the OpenAI format: set the base URL and your Luger key.
What does the API cost?
Per token, from the same balance as the chat. From $0.40 per 1M input tokens. No subscription.
Is my code stored?
Chat history lives in your account and deletes in one click; a private chat is not saved at all. Conversations are not used to train models.