← Leaderboard
yuxinlu1/gemma-4-12b-opus-reasoning

Gemma 4 12B Opus Reasoning Q4

llama.cpp 12B Reasoning
Throughput
17.3t/s@ 32k
Engine
llama.cpp
Parameters
12B
Benchmarked
2026-06-27

Context ladder

Throughput at each benched context window (single measurement).

Context KV Throughput
@ 32k peak golden q8_0 17.3t/s

Golden profile

gemma4-12b-opus-reasoning-q4

Capabilities

testingreasoningllamacppgoldengguf

Why we run it

Golden fleet target — auto-scaffolded from recipe gemma4-12b-opus-reasoning-q4.

Bench notes

golden 32k/q8_0 @ 17.3 tok/s — fill~14745 — bench-agent-v2 — tool_ok=True

Links

Benchmarked 2026-06-27
SparkBench · GB10 · single node