Muna logomuna

Benchmarks

We measure Muna-compiled language and image models on artifact size, cold starts, and inference performance. Every number is derived from raw result files you can download, with the run configuration published alongside it.

Methodology · Benchmark your model

FLUX.2 Klein 9B

report →
@black-forest-labs/flux-2-klein-9bimageNVIDIA B200 · August 2026
compiled code size
16MB
compiled code size
artifact pull
1.6GB
artifact pull
median cold-start time to first image
5.0s
median cold-start time to first image
median generation time · 1024×1024
426ms
median generation time · 1024×1024

Gemma 4 26B A4B

report →
@google/gemma-4-26b-a4b-itchatNVIDIA B200 · May 2026 · vs SGLang
compiled code size
96MB
compiled code size
artifact pull, vs 25.8GB
760MB
artifact pull, vs 25.8GB
CTTFT @p90 · Muna (bare metal)
6.3s
CTTFT @p90 · Muna (bare metal)
lower median CTTFT vs SGLang (Modal)
18×
lower median CTTFT vs SGLang (Modal)