Chip comparison
Google TPU v5p vs Google TPU v6 (Trillium)
Side-by-side specs
Straight from each vendor's datasheet. Winner in bold: higher throughput, memory, or transistor count; lower power draw or price. Sparse (2:4) throughput roughly doubles the FP8/INT8 numbers shown.
| Field | Google TPU v5p | Google TPU v6 (Trillium) |
|---|---|---|
| TDP | 300 W | 200 W |
| Memory | 95 GB HBM2e | 32 GB HBM3 |
| Memory bandwidth | 2,765 GB/s | 1,600 GB/s |
| BF16 (dense) | 459 TFLOPS | 918 TFLOPS |
| FP8 (dense) | — | 1,836 TFLOPS |
| INT8 (dense) | 918 TOPS | 1,836 TOPS |
| Form factor | OAM (per-chip in v5p pod) | OAM (Trillium pod, up to 256 chips) |
| Announced | 2023-12-06 | 2024-05-14 |
| Released | 2024-01-01 | 2024-12-01 |
Robots running Google TPU v5p
No robots publicly running it yet.
Robots running Google TPU v6 (Trillium)
No robots publicly running it yet.
Data centers with Google TPU v5p
Data centers with Google TPU v6 (Trillium)
Common questions
Google TPU v5p vs Google TPU v6 (Trillium): which is faster for training?
Google TPU v6 (Trillium) has 4.00x the dense FP16/BF16 throughput of the other (Google TPU v5p: 459 TFLOPS; Google TPU v6 (Trillium): 1,836 TFLOPS). Real-world training speed also depends on memory bandwidth, interconnect, and software stack: see the spec diff for those inputs.
Google TPU v5p vs Google TPU v6 (Trillium): which has more memory?
Google TPU v5p carries more HBM (Google TPU v5p: 95 GB HBM2e; Google TPU v6 (Trillium): 32 GB HBM3). Memory capacity sets the largest model that fits on one chip without partitioning; memory bandwidth sets inference throughput.
Google TPU v5p vs Google TPU v6 (Trillium): which is more power-efficient?
Dense FP16/BF16 TFLOPS per watt: Google TPU v5p 1.53 (459 ÷ 300 W); Google TPU v6 (Trillium) 9.18 (1,836 ÷ 200 W). Google TPU v6 (Trillium) wins at the die level; system-level efficiency also depends on cooling and interconnect.
See also: every chip comparison · Google TPU v5p full page · Google TPU v6 (Trillium) full page.