Feature Overview
- ๐จ Brand new UI (some pages are still being updated)
- ๐ Multi-language support (to be improved)
- ๐จ Added Midjourney-Proxy(Plus) API support
- ๐ฐ Supports online top-up, can be set in system settings:
- EasyPay
- ๐ Supports checking usage by key:
- With the project neko-api-key-tool, you can check usage by key
- ๐ Pagination supports selecting the number of items per page
- ๐ Compatible with the original One API database, can directly use the original database (one-api.db)
- ๐ต Supports per-call model billing, can be set in System Settings - Operation Settings
- โ๏ธ Supports random channel weighting
- ๐ Data dashboard (dashboard)
- ๐ Can set which models a token can call
- ๐ค Supports Telegram OAuth login:
- System Settings - Configure Login/Register - Allow login via Telegram
- Send command /setdomain to @Botfather
- Select your bot, then enter http(s)://your-website-address/login
- Telegram Bot name is the bot username without the @
- ๐ต Added Suno API support
- ๐ Supports Rerank models, currently compatible with Cohere and Jina, can be integrated with Dify
- โก OpenAI Realtime API - Supports OpenAI's Realtime API, supports Azure channels
- Supports entering the chat interface via /chat2link route
- ๐ง Supports setting reasoning effort via model name suffix:
- OpenAI o-series models
- Add suffix
-highfor high reasoning effort (e.g.:o3-mini-high) - Add suffix
-mediumfor medium reasoning effort (e.g.:o3-mini-medium) - Add suffix
-lowfor low reasoning effort (e.g.:o3-mini-low)
- Add suffix
- Claude reasoning models
- Add suffix
-thinkingto enable reasoning mode (e.g.:claude-3-7-sonnet-20250219-thinking)
- Add suffix
- OpenAI o-series models
- ๐ Reasoning-to-content: supports setting
thinking_to_contentin Channel - Edit - Channel Extra Settings, default isfalse. When enabled, reasoning contentreasoning_contentwill be converted to<think>tag and appended to the content in the response. - ๐ Model rate limiting: supports setting model rate limits in System Settings - Rate Limit Settings, supports setting total request limit and successful request limit
- ๐ฐ Cache billing support: after enabling, you can bill according to the set ratio when cache hits occur:
- Set prompt cache ratio option in System Settings - Operation Settings
- Set prompt cache ratio in channels, range 0-1, e.g. 0.5 means 50% billing on cache hit
- Supported channels:
- OpenAI
- Azure
- DeepSeek
- Claude