Features
Upload PDF, DOCX, TXT, Markdown
Automatic chunking and embedding
Cited answers linking to source documents
REST API for chat, history, and search
Separate knowledge bases with per-project API keys
System prompt and model selection
Retrieval tuning without ML experience
Analytics on queries and citation frequency
Live preview of assistant before integration
One endpoint for retrieval, generation, and citations
Dashboard for managing knowledge bases, documents, and API keys
Conversation management for testing and review
Token streaming for instant feel
Email-based password reset with one-time link
OpenAI-compatible API for model serving
Deploy 500+ open-source models
Write custom GPU kernels with Mojo
Zero dependency on PyTorch, CUDA, or ROCm
Single container under 700MB for self-hosted
Paged KV cache for memory efficiency
Quantization (bfloat16, float32)
Multi-node distributed inference
Model customization via PyTorch-like API
Hardware-agnostic (NVIDIA, AMD, Apple Silicon)
Mixture-of-Experts serving
max benchmark tool adapted from vLLM
MiniMax M3 open weights support
SOTA MoE serving on Modular Cloud
