Note: Announcing preview access to Trillium: Our sixth generation and most advanced TPU to date.
Cloud TPUs are purpose-built AI accelerators, which are fully optimized for a diverse range of AI training, tuning, and inference workloads.
Learn which Cloud TPU products works best for your unique project needs.
Note: Contact our sales team for access to Cloud TPU v5p, our most powerful and scalable Cloud TPU to date, now generally available.
Cloud TPU pricing varies by product, deployment model & region. Charges for Cloud TPU accrue while a TPU node is in a READY state. Customers receive a bill at the end of each billing cycle that lists usage and charges for that billing cycle. Prices displayed here are per chip-hour in USD.
Cloud TPU Version (per chip-hour) | Region | Location | On Demand (USD) (per hour or per month) | 1-year Commitment (USD) (per hour or per month) | 3-year Commitment (USD) (per hour or per month) |
---|---|---|---|---|---|
Trillium | us-east1 | South Carolina | $2.7 | $1.89 | $1.22 |
us-east5 | Ohio | $2.7 | $1.89 | $1.22 | |
europe-west4 | Amsterdam | $2.97 | $2.08 | $1.34 | |
asia-northeast1 | Tokyo | $3.24 | $2.27 | $1.46 | |
TPU v5p | us-east5 | Columbus | $4.2 | $2.94 | $1.89 |
us-east1 | South Carolina | $4.2 | $2.94 | $1.89 | |
TPU v5e | us-central1 | Iowa | $1.2 | $0.84 | $0.54 |
us-east5 | Columbus | $1.2 | $0.84 | $0.54 | |
us-south1 | Dallas | $1.416 | $0.9912 | $0.6372 | |
us-west1 | Oregon | $1.2 | $0.84 | $0.54 | |
us-west4 | Las Vegas | $1.2 | $0.84 | $0.54 | |
europe-west1 | Belgium | $1.32132 | $0.924924 | $0.594594 | |
europe-west4 | Netherlands | $1.56 | $1.092 | $0.702 | |
asia-southeast1 | Singapore | $1.56 | $1.092 | $0.702 | |
TPU v4 pod | us-central2 | Oklahoma | $3.22 | $2.0286 | $1.449 |
TPU v3 pod | europe-west4 | Netherlands | $2 | $1.26 | $0.9 |
TPU v3 device | europe-west4 | Netherlands | $2.2 | $1.386 | $0.99 |
TPU v2 pod | us-central1 | Iowa | $1.5 | $0.945 | $0.675 |
europe-west4 | Netherlands | $1.5 | $0.945 | $0.675 | |
TPU v2 device | asia-east1 | Taiwan | $1.305 | $0.82215 | $0.58725 |
europe-west4 | Netherlands | $1.2375 | $0.779625 | $0.556875 |
Spot prices are dynamic and can change up to once every 30 days. For more information, see the Spot VMs documentation.
Type | Billing | Best fit for: |
---|---|---|
3-year commitment (3Y CUD) | Monthly, based on reserved quota | ML users who need a more consistent access to 512+ cores of capacity compared to On Demand or Spot options |
1-year commitment (1Y CUD) | Monthly, based on reserved quota | ML users who need a more consistent access to 32-512 cores of capacity compared to On Demand or Spot options |
On Demand | Hourly, based on actual usage | ML users who want to run some short-term experiments or benchmarks |
Spot/Preemptible | Hourly, based on actual usage | ML users who want to run batch / fault-tolerant workloads |
Pricing shown above is per chip-hour. A single TPU Virtual Machine (VM) can have multiple chips and at least 2 cores. Billing in the Google Cloud console is displayed in VM-hours (for example, the on-demand price for a single Cloud TPU v4 host, which includes four TPU v4 chips and one VM, is displayed as $12.88 per hour). Usage data in the Google Cloud console is also measured in VM-hours.
Estimate the cost of using Cloud TPU with the Compute Engine pricing calculator, choose “Cloud TPU” on the top bar.
New customers get $300 in free credits to spend on Google Cloud. Get started now.
If you are a researcher, student, tinkerer, artist, or entrepreneur, consider applying to the TPU Research Cloud program. TRC members are granted free access to a large cluster of Cloud TPUs, and they share their work with the world via peer-reviewed publications, open source code, blog posts, videos, and other media. (Here are examples of TRC-supported publications.)
Apply to accelerate your research today!