Instant $10 Free Credits
Sign up via our referral link and get $10 in GPU credits instantly. That's ~55 hours on RTX 3080, ~45 hours on RTX 3090, or ~22 hours on RTX 4090 — completely free. No credit card required for the trial.
Search 500,000+ Hugging Face models. One-click deploy to RunPod serverless GPUs. Auto-scaling from 0 to 1000+ workers. Pay per second. Zero infrastructure.
Curated models optimized for RunPod serverless — one-click deploy, pre-tested configs.
Search 500,000+ Hugging Face models. Filter by task, size, license. Deploy to RunPod instantly.
Choose the right GPU for your model size and budget. All serverless, per-second billing. Spot saves 50-70%.
Sign up via our referral link and get $10 in GPU credits instantly. That's ~55 hours on RTX 3080, ~45 hours on RTX 3090, or ~22 hours on RTX 4090 — completely free. No credit card required for the trial.
Every person you refer earns you 20% of their spend — forever. If they spend $500/mo on GPUs, you get $100/mo passive income. 10 referrals at $500 = $5,000/mo. No caps. No expiry. Track it all at your referral dashboard.
Access 100,000+ GPUs across 30+ data centers globally. RTX 3080 ($0.17/hr) to H100 ($4.50/hr). Sub-second cold starts. Per-second billing. Auto-scale from 0 to 1000+ workers. No reserved instances needed.
Deploy any HF model ID directly. Auto-detects architecture, quantization (AWQ, GPTQ, GGUF), and required VRAM. OpenAI-compatible API endpoints. Built-in Flash Attention, vLLM, TGI, and custom Docker support.
Bid on spare capacity. RTX 3090 at $0.11/hr (vs $0.22), A100 40GB at $0.55/hr (vs $1.10). Perfect for batch inference, fine-tuning, and fault-tolerant workloads. Automatic fallback to on-demand.
SOC 2 Type II, HIPAA-ready, GDPR compliant. Private networking, VPC peering, custom VPCs. End-to-end encryption. Team workspaces with RBAC. Audit logs. Dedicated support SLAs.
Launch LoRA/QLoRA fine-tunes on any HF model. Multi-node distributed training with NCCL. Pre-configured Axolotl, Unsloth, HuggingFace Trainer templates. Checkpoint to HF Hub automatically.
Deploy inference endpoints at edge locations worldwide. Cloudflare Workers integration. <100ms latency to 95% of internet users. Custom domains, SSL, rate limiting, auth built-in.
Log into your referral dashboard to get your personal link, track clicks, signups, conversions, and pending/paid commissions in real-time. Share your link on GitHub, Discord, Twitter, blogs, YouTube — every deploy earns you back.
Open Referral Dashboard →Deploy Llama 3.1, Mistral, Qwen, Whisper, or any custom fine-tune. Your $10 free credits cover your first workloads. Then every referral you bring in pays for your GPU bill — indefinitely.
From Hugging Face model ID to production API in 3 steps.
Search any HF model ID or browse curated lists. We auto-detect architecture, quantization, and required VRAM.Browse Models →
Choose from RTX 3080 to H100. We recommend minimum VRAM for your model. Spot instances save 50-70%.
View GPUs →One click deploys a serverless worker. Auto-scales from 0 to 1000+ replicas. OpenAI-compatible API endpoint ready.Deploy Now →