LithoCore AI Developer Documentation
Welcome to the LithoCore AI REST & Streaming API. LithoCore AI provides a unified, hyper-fast inference gateway to execute autonomous agents, perform multimodal document RAG, and stream 48kHz neural voice with sub-30ms latency.
https://api.lithocoreai.com/v1
Authentication
The LithoCore API uses Bearer API keys for request authorization. You can obtain a free sandbox key from your Console Playground.
Official Client SDKs
Install our native client libraries for seamless type safety, automated exponential backoff, and WebSocket streaming.
pip install lithocore-ai
npm install @lithocore/sdk
/v1/chat/completions
Generate streaming completions, tool call invocations, or multi-turn dialogues using LithoCore's foundational models.
{
"model": "litho-4-omni",
"messages": [
{"role": "system", "content": "You are an enterprise AI data assistant."},
{"role": "user", "content": "Summarize latency bottlenecks across our AWS Aurora cluster."}
],
"temperature": 0.2,
"stream": true
}
{
"id": "chatcmpl_9a04f",
"object": "chat.completion",
"created": 1774392810,
"model": "litho-4-omni",
"choices": [{
"index": 0,
"message": {
"role": "assistant",
"content": "Analysis complete. 3 index scans detected with over 40ms IOPS wait time on table order_ledger..."
},
"finish_reason": "stop"
}],
"usage": {
"prompt_tokens": 42,
"completion_tokens": 128,
"total_tokens": 170
}
}
AWS VPC & PrivateLink Integration
For enterprise customers with stringent compliance obligations, LithoCore AI supports AWS PrivateLink. This allows your Amazon ECS, EKS, or EC2 workloads to query our inference endpoints entirely over the AWS internal backbone without routing traffic over the public internet.