Features
OpenAI-compatible chat completions API
Text generation with automatic crawling of linked content
Image analysis and vision (multimodal)
Speech-to-text (8x cheaper than Google Cloud)
Text-to-speech with Kokoro TTS integration
Unified embeddings for text, images, and code
Prebuilt tools: Domain Name Generator, AI Text Editor, Prompt Optimizer, Image Captioning
Self-hosting option for full data control
Sub-second response times
Privacy-first: no unnecessary data retention
Streaming support for chat completions
Summarization API with max length parameter
Audio-in audio-out via Gemma 4 Multimodal Audio API
Qwen3.5-4B with reasoning and thinking support
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
