VAST.AI ALTERNATIVE
Like Vast.ai, but with predictable supply.
Vast.ai is a marketplace — supply varies, hosts vary, reliability varies. Runcrate is a single operator with consistent per-GPU pricing, multi-region availability, and a 200+ model inference API on the same platform. Cheaper than Vast on most flagship GPUs.
COMPARISON
Runcrate vs Vast.ai.
| Feature | Runcrate | Vast.ai |
|---|---|---|
| H100 PCIe | $1.40/hr (consistent) | $1.20-3.00/hr (varies) |
| Hardware reliability | First-party datacenter | Third-party hosts (varies) |
| Inference API | 200+ models built-in | Not offered |
| Multi-GPU clusters | Up to 8x NVLink | Marketplace-dependent |
| SLAs | 99.9% uptime | Best-effort (host-dependent) |
| Per-second billing | Yes | Per-second on most hosts |
GPU PRICING
GPU pricing comparison.
| Model | Provider | Price | Detail |
|---|---|---|---|
| deepseek-ai/DeepSeek-V3.2 | DeepSeek | $0.27 / 1M | Reasoning, code, 128K ctx |
| anthropic/claude-4-sonnet | Anthropic | $3 / 1M in, $15 / 1M out | Top-tier reasoning |
| meta-llama/Llama-4-Scout | Meta | $0.20 / 1M | Open weights, multilingual |
| Qwen/Qwen3-Max | Alibaba | $0.30 / 1M | 30+ languages, 128K ctx |
| openai/whisper-large-v3 | OpenAI | $0.02 / min | Speech-to-text, 100+ langs |
| black-forest-labs/FLUX.1-pro | Black Forest Labs | $0.04 / image | Photorealistic |
WHY SWITCH
Why teams switch to Runcrate.
200+ models, one API key
Chat, code, image, video, audio, embeddings, vision — all under a single OpenAI-compatible endpoint with per-token / per-image / per-second billing.
OpenAI-compatible drop-in
Swap the base URL and your existing OpenAI SDK code keeps working. No custom client library, no rewrite, no lock-in.
Inference + GPU rentals
When the API isn't enough, rent a dedicated H100, H200, or B200 from the same account — same billing, same dashboard, no separate vendor.
Per-second billing, no minimums
Pay only for what you use. No hourly bucketing, no commitment, no idle charges. Prepaid credits never expire.
GET STARTED
Try it now.
import Runcrate from "@runcrate/sdk";
const rc = new Runcrate({ apiKey: "rc_live_YOUR_API_KEY" });
// Spin up a dedicated H100 SXM in 60 seconds
const instance = await rc.instances.create({
gpu: "h100-sxm",
region: "auto",
image: "runcrate/vllm:latest",
});
console.log(`SSH: ssh root@${instance.host}`);FAQ