Learn, Build, Deploy

Complete REST API documentation. GPU instances, inference endpoints, volumes, SSH keys, billing — every endpoint with request/response schemas.

Create an account, generate an API key, launch your first GPU instance, and SSH in — all in under 5 minutes. Step-by-step walkthrough for new users.

JWT-based auth with access and refresh tokens. API key authentication for programmatic access. Rate limits, token rotation, and security best practices.

Deploy Kimi K2.5 on Ubuntu

Complete guide to running Moonshot AI's Kimi K2.5 (1T parameter MoE) locally on Ubuntu with Ollama, llama.cpp, or vLLM. Covers quantization, hardware requirements, and benchmarks.

Fine-tune LLaMA 3 with LoRA

Step-by-step guide to fine-tuning Meta LLaMA 3 on your custom dataset using LoRA adapters on a single A100 or H100 GPU. Includes dataset preparation and evaluation.

Multi-GPU Training with DeepSpeed

Set up distributed training across multiple GPU nodes using DeepSpeed ZeRO-3. Configure InfiniBand networking, NCCL, and checkpoint saving for large model training.

Deploy vLLM for Production Inference

Run vLLM as a production inference server on gpuLabs. OpenAI-compatible API, continuous batching, PagedAttention, and auto-scaling configuration.

Build a RAG Pipeline on GPU Cloud

End-to-end retrieval-augmented generation: embed documents with BGE, index in a vector database, and serve answers with a fine-tuned LLM on dedicated GPU.

Run Stable Diffusion XL at Scale

Batch image generation with SDXL on high-VRAM GPUs. API setup, prompt engineering, ControlNet integration, and scaling to thousands of images per hour.

Create support tickets directly from your dashboard. Our engineering team responds within 2 hours during business hours. Priority support for enterprise accounts.