RTX 3090 Qwen3-8B Q4_K token-gen @4K ctx
115.3 tok/s TG batch-1 @4K ctx
Fields
id
benchmarks/comm-rtx-3090-qwen3-8b-q4-tg-4kMetric
decode_tok_s
Value
115.3
Unit
tok/s TG batch-1 @4K ctx
Model
Qwen3-8B
Methodology
llama.cpp llama-bench, Q4_K GGUF, CUDA, batch-1 decode @4K context (Hardware Corner measured table)
Measured by
third_party
Thermal state
unknown
notes
88 t/s @16K, 68 t/s @32K. Reference card for the class. Also 3090 Ti +9% (123.7).
Notes
88 t/s @16K, 68 t/s @32K. Reference card for the class. Also 3090 Ti +9% (123.7).