live eventEidos is moving itself to 90%-cheaper silicon without losing intelligence. Watch.follow on LinkedIn →
mission: 90% local authorship — waiting for first eventhow?
← charts + scores

llama3.2:1b on A6000, A100, H100 — 2026-04-17

2026-04-17T22:08:38.000Z 2026-04-17T22:09:31.296Z
tweet →

On 2026-04-17, llama3.2:1b ran across A6000, A100, H100. H100 finished first at 590.0 tok/s, 12× faster than A100. The per-million-token cost: $1.17 (H100) vs $4.50 (A100) — the surprise: the headline GPU H100 was also the cheapest per-million tokens, coming in 3.8× less than A100.

Podium

H100
1st
thunder-h100
peak tok/s
590.0
avg tok/s
590.0
$ / 1M tok
$1.17
A6000
2nd
thunder-a6000
peak tok/s
72.8
avg tok/s
72.8
$ / 1M tok
$1.34
A100
3rd
thunder-a100
peak tok/s
48.1
avg tok/s
48.1
$ / 1M tok
$4.50

pit wallH100 just posted 193 tok/s on llama3.1:8b — 13.7× faster than the A100 lane (1428h ago).

llama3.2:1b on A6000, A100, H100 — 2026-04-17 — Eidos Live · Crucible