跳到主要内容
One API · 170+ Backends · Real Devices

Collapse language into real creation

LiMa is a quantum-routed AI cloud for smart hardware. Talk, draw, write, and animate through a single OpenAI-compatible endpoint.

Partners

接入 170+ AI 后端

与全球领先模型与平台无缝协作。

GPT-4o
Claude
DeepSeek
Groq
NVIDIA
OpenRouter
Cloudflare
Google AI
Gemini
Mistral
SiliconFlow
Zhipu
Baidu
Tencent
Volcengine
Aliyun
Llama
LongCat
Cohere
Replicate
Together AI
FAQ

Common questions

Which AI models does LiMa support?+

LiMa connects to 170+ backends across GPT-4o, Claude, DeepSeek, Groq, NVIDIA, OpenRouter, Cloudflare, Gemini, Mistral, SiliconFlow, Zhipu, Baidu, Tencent, Volcengine, Alibaba Cloud and more.

What hardware can I connect?+

ESP32-S3 / ESP32-C3 based boards. We provide reference designs for an AI drawing machine, AI writing machine, and 2D digital human.

Is my data secure?+

LiMa uses TLS transport, device remote attestation, signed OTA updates, A/B rollback, and access whitelists for end-to-end security.

How do I deploy LiMa?+

One-command Docker, VPS, or local development. Production uses nginx + uvicorn with automated deployment scripts.

How is billing calculated?+

Usage is metered by tokens and requests. The console shows 30-day statistics. The free tier includes 50 chats per day.

Do you support private deployment?+

Yes. Enterprise plans include on-premises deployment, SLA, custom hardware adaptation, and dedicated support.

Start building with LiMa today

Free tier available. Upgrade when you scale.