Community submission

Performance record · MoE

GPT OSS 120B

MXFP4 running on AMD Radeon PRO W6800X Duo + AMD Radeon Pro Vega II with macOS 26.7.3 Tahoe

PERFORMANCE / TOKENS PER SECONDRDNA 2 + GCN / Vega
Prompt processing1488.43tok/s
#49 in index
Token generation70.73tok/s
#84 in index

Run fingerprint

Enough context to make the number useful.

QuantizationMXFP4Exact submitted variant
GPU architectureRDNA 2 + GCN / VegaMetal backend
Test context512prompt tokens
Model familyMoESigned ToshLLM app · amd-gpu · Metal
ToshLLM versionv0.88.0App version used to capture and sign this run
01

Standard workload

Prompt processing and generation are reported separately so memory bandwidth and decode behavior stay visible.

02

Configuration attached

Quantization, GPU family, VRAM and context travel with the result instead of being hidden behind one score.

03

Provenance preserved

This result remains explicitly labeled as a community submission.