MN API is an aggregation platform that relays API requests across multiple upstream AI relay providers, removing the need to pick a single vendor. It curates upstream relays by cost-effectiveness and model coverage, with most group pricing matching the source relay. The service supports high concurrency and performs real-time quality monitoring on each connected upstream.
Supported models and groups include:
- Grok 3: Grok-3, Grok-3-deepsearch, Grok-3-reasoner, Grok-4.
- DeepSeek: full fast (164k context), full (128k context), and az (64k context) variants.
- Azure: mixed and uncensored options.
- OpenAI: official keys and per-request billing for models like GPT-4o, O3-mini, and image generation.
- Gemini: 2.5 series, exp channel, and per-request billing including 2.5 Pro-all and 2.5 Pro-preview.
- Claude: official key, discounted official, and code channel options.
- Others: QwQ-32B, free OpenRouter models (including Kimi K2).
Pricing is tiered by model group, with per-request billing available on some groups. The platform offers mainland China acceleration and supports sign-up via Gmail, QQ email, GitHub, LinuxDo, and Google. New users receive a $1 balance on sign-up, with no minimum top-up requirement. A free Telegram chat bot supports GPT-4o, GPT-4o-mini, Gemini-2.5-flash, and Gemini-2.5-pro.
