NVIDIA RTX 5090 llama.cpp: Qwen3.5 35B (MXFP4) prompt processing at 256k context
2003.7 tok/s (pp256k, per-request batch-1)
Fields
id
benchmarks/hc26-5090-qwen35-35b-mxfp4-256k-ppMetric
prefill_tok_s
Value
2,003.7
Unit
tok/s (pp256k, per-request batch-1)
Model
Qwen3.5 35B (MXFP4)
Methodology
THIRD-PARTY MEASURED (Hardware Corner GPU LLM Benchmarks page, tables dated March 2026). llama.cpp llama-bench, GGUF, single GPU, batch-1. NVIDIA RTX 5090. Qwen3.5 35B (MXFP4). Number is the 256k-context column of the Prompt Processing table: context depth the run was measured at, not a fixed pp512/tg128. Per-request, NOT aggregate.
Measured by
third_party
Reproducible
yes