# GPUniq (https://gpuniq.com)

> GPUniq is a GPU meta-cloud and rental marketplace for AI, machine learning, rendering and scientific computing. 74 GPU models with public price/availability analytics, 96 models and 15,956 GPUs live in the network right now, from $0.04/hr. Automatic failover, snapshots, per-second GPU billing, OpenAI- and Anthropic-compatible LLM API on a single balance. Data on this page: GPUniq live marketplace analytics, 2026-09-09.

## Links

- [Main site](https://gpuniq.com): GPU rental marketplace
- [GPU statistics hub](https://gpuniq.com/statistics): live pricing & availability for all 74 GPU models
- [GPU Cloud Price Index](https://gpuniq.com/gpu-price-index): monthly median $/hr per model, with methodology
- [VRAM calculator](https://gpuniq.com/vram-calculator): which LLMs fit which GPU
- [Search GPUs](https://gpuniq.com/search): filter by VRAM, region, price; deploy in one click
- [Blog](https://gpuniq.com/blog): practical guides on GPU cloud economics
- [Updates / changelog](https://gpuniq.com/updates)
- [API docs](https://docs.gpuniq.com): REST API, OpenAI-compatible LLM endpoint, MCP server
- [OpenAPI spec](https://api.gpuniq.com/openapi.json)
- [Full GPU catalog for LLMs](https://gpuniq.com/llms-full.txt)
- [Statistics pages index for LLMs](https://gpuniq.com/llms-statistics.txt)

## What GPUniq does

- Rents 74+ GPU models — data-center (B200, H200, H100, A100, L40S), workstation (RTX 6000 Ada, RTX A6000) and consumer (RTX 5090/4090/3090 and more) — by the hour with per-second billing.
- Automatically picks the optimal GPU for a task and budget; takes snapshots and migrates workloads to a new machine on failure (automatic failover).
- Tracks live prices and availability across the whole network; all analytics are public, no login required.
- Serves an LLM API (OpenAI-compatible `/v1/openai/*`, native Anthropic `/v1/messages`, MCP) billed from the same balance as GPUs.

## Popular GPUs — live rental prices (GPUniq marketplace data, 2026-09-09)

| GPU | VRAM | Min $/hr | Median $/hr | Available now |
|-----|------|----------|-------------|---------------|
| NVIDIA B200 | 192GB HBM3e | $8.01 | $11.32 | 43 |
| NVIDIA H200 | 141GB HBM3e | $5.19 | $5.98 | 82 |
| NVIDIA H100 | 80GB HBM3 | $2.61 | $4.02 | 102 |
| NVIDIA A100 | 40–80GB HBM2e | $0.64 | $1.43 | 117 |
| NVIDIA GeForce RTX 5090 | 32GB GDDR7 | $0.50 | $0.82 | 325 |
| NVIDIA GeForce RTX 4090 | 24GB GDDR6X | $0.40 | $0.66 | 265 |
| NVIDIA GeForce RTX 3090 | 24GB GDDR6X | $0.17 | $0.25 | 152 |
| NVIDIA L40S | 48GB GDDR6 | $0.91 | $1.11 | 19 |
| NVIDIA RTX 6000 Ada Generation | 48GB GDDR6 | $0.68 | $1.09 | 28 |
| NVIDIA GeForce RTX 3060 | 12GB GDDR6 | $0.06 | $0.10 | 163 |

Full catalog with prices for all 74 models: https://gpuniq.com/llms-full.txt

## Contact

- Site: https://gpuniq.com
- API: https://api.gpuniq.com