NVIDIA A100 80GB (SXM4)
Compare NVIDIA A100 80GB (SXM4) against another machine in the Arena
Reference: Ampere (microarchitecture) on Wikipedia
- Class
- GPU
- Memory
- 80 GB HBM2e
- Runs (Q4_K_M)
- —
- Bandwidth
- 2039 GB/s
- TDP
- 400 W
- Released
- 2020-11-01
- Price (US)
- $16950 used · $19789 new as of 2026-08
Sourcing and disambiguation notes
The spec figures above (2039 GB/s, 400W) are for the SXM4 module, which needs an HGX baseboard and is rarely resold on its own. Most 'used A100 80GB' listings, including the price recorded here, are almost certainly the PCIe card (~1935 GB/s, ~300W), so treat the price as a PCIe street price and the bandwidth and TDP as SXM4 spec until re-verified. The source does not state which form factor its figure covers; the higher, SXM-typical price was assigned here on the inference that SXM4 commands a premium over the separately priced a100-80gb-pcie entry, which is reasoning rather than a confirmed spec match.
Specification sources
- https://gpudojo.com/a100-80gb — press, 2026-08-22
- https://cputronic.com/en/gpu/nvidia-a100-sxm4-80-gb — press, 2026-08-22
Measurements
Decode is token generation — the speed you feel while an answer streams. Prefill is prompt processing — the wait before it starts. Why bandwidth predicts decode speed.
No records yet for this hardware. Know of a published benchmark on it? Submit the link.
What verified, single-source and estimated mean, and the same rows with every filter and sort in the benchmarks explorer.
Speed over time
What it can run
Fit is arithmetic, not a measurement — how it is computed. Against 80 GB; models that fit are listed largest first.
No modeled quant fits in 80 GB.