Gaudi 3 specs
Intel · Shipping now · first available Sep 2024 (estimated). Dense compute, memory, bandwidth and power from the vendor's documents; each precision stays separate.
All accelerators · Compare with TPU v6e (Trillium)
Specifications
| Dense BF16 · FP32 accumulate | 1,678 TFLOPS |
|---|---|
| Dense FP8 · FP32 accumulate | 1,678 TFLOPS |
| Dense FP4 · FP32 accumulate | Not published |
| Memory | 128 GB · HBM |
| Memory bandwidth | 3.7 TB/s |
| Maximum chip power | 900 W |
| Interconnect | 24×200 GbE RoCE |
| Process node | TSMC 5nm |
| First availability | Sep 2024est — estimated / secondary-source figure |
| Availability | Shipping |
| Est. purchase price | ~$15,625est — estimated / secondary-source figureEight-accelerator kit list price ÷ 8, baseboard included |
Priced aggressively; FP8 and BF16 share the same MME throughput.
Dense tensor rates with FP32 accumulation, never sparse or FP16-accumulate headline figures. Memory is in decimal GB. The est marker flags a value its vendor-document review did not confirm.
Rent it
No tracked rental quote. The site tracks providers' published list prices; this part may still be rentable under terms it does not record.
Value ratios
| $ per dense BF16 PFLOP-hour | No on-demand quote |
|---|---|
| Dense BF16 TFLOPS per watt | 1.86 |
| Bandwidth bytes per 1,000 BF16 FLOP | 2.21 |
| $ per dense FP8 PFLOP-hour | No on-demand quote |
| Dense FP8 TFLOPS per watt | 1.86 |
| Bandwidth bytes per 1,000 FP8 FLOP | 2.21 |
| $ per GB of memory per hour | No on-demand quote |
Prices divide the cheapest on-demand list price; theoretical ratios do not measure application speed.
Sources
Specifications reviewed . Intel's July 2025 paper gives 1,678 TFLOPS BF16/FP8 MME at 900 W; its 2024 Hot Chips deck said 1,835.
Purchase price: Intel Computex 2024 keynote (archived) ↗ (2024-06-04): “A kit including eight Intel Gaudi 3 accelerators with a UBB will list at $125,000”