Google Cloud
Google Cloud sells Gemini and partner models per token through Vertex AI and rents GPU machines by the hour on Compute Engine.
- Website: https://cloud.google.com
- Headquarters: Mountain View, United States
- Kind: Hyperscaler
Google Cloud sells 0 of the compared models per token and prices 5 kinds of GPU for hosting your own.
GPUs for hosting your own
| GPU | From, per GPU-hour | Cheapest instance | GPUs in it |
|---|---|---|---|
| L4 | $0.7068 | g2-standard-4 | 1 × 24 GB |
| A100 | $3.48 | a2-megagpu-16g | 16 × 40 GB |
| RTX PRO 6000 | $4.50 | g4-standard-48 | 1 × 96 GB |
| H200 | $10.60 | a3-ultragpu-8g | 8 × 141 GB |
| H100 | $11.06 | a3-highgpu-8g | 8 × 80 GB |
Where it runs
1 region in 1 country.
| Country | Regions | Per-token models | GPU kinds |
|---|---|---|---|
| United States | us-central1 | — | 6 |
Where the prices come from
- Per-token prices: cloud.google.com/vertex-ai/generative-ai/pricing
- GPU servers: cloud.google.com/products/compute/pricing/accelerator-optimized
- Data sent out (egress): cloud.google.com/vpc/network-pricing
Sources
List prices collected 4 November 2026, refreshed every three hours. No negotiated rates or taxes.