qwen2.5:1.5b on A100, H100 — 2026-04-18
2026-04-18T06:58:32.000Z → 2026-04-18T06:58:55.107Z
On 2026-04-18, qwen2.5:1.5b ran across A100, H100. H100 finished first at 406.0 tok/s, 19× faster than A100. The per-million-token cost: $1.70 (H100) vs $10.17 (A100) — the surprise: the headline GPU H100 was also the cheapest per-million tokens, coming in 6.0× less than A100.
Podium
H100
1st
thunder-h100
- peak tok/s
- 406.0
- avg tok/s
- 406.0
- $ / 1M tok
- $1.70
A100
2nd
thunder-a100
- peak tok/s
- 21.3
- avg tok/s
- 21.3
- $ / 1M tok
- $10.17