Features
200+ models via single serverless API
LLMs: Deepseek V4 Pro, Qwen3.7-Max, Kimi K2.7, Gemma 4, etc.
Image generation and editing models
Audio models (speech recognition, TTS)
Vision / multimodal models (Qwen-VL, etc.)
Agent Sandbox for secure code execution
Dedicated GPU instances (H200, H100) per-second billing
Serverless GPU (pay-per-execution batch jobs)
Low latency ~200ms, 99.5% uptime
Token-based billing with cache discounts
Batch inference at 50% discount
NVLink and GPUDirect RDMA clusters
Integrates with Harbor agent evaluations
API key management and usage metrics
Access 10+ AI models: OpenAI, Claude, Gemini, Grok-3, DeepSeek, Qwen, etc.
1,000+ pre-built AI apps for content, image, video, voice, automation
No-code AI app builder for custom text/image/workflow apps
Drag-and-drop workflow builder with nodes and chains
Batch processing using table data sources (hundreds of records)
Auto Agent builder for autonomous task resolution
Train chatbots on own data (brand voice integration)
Multi-modal generation: text, image, video, voice
Vector storage for document embeddings (up to 8GB on Premium)
Document upload limit scaling from 50 to 2000 documents
Credit-based system with free tier (30 credits/day)
Web-based platform (desktop only, no mobile support)
On-premises deployment for enterprise
API integration on paid tiers
Third-party integration on paid tiers