RTX 5090 Qwen3-8B Q4_K_XL token-gen @4K
185.9 tok/s TG batch-1 @4K ctx
Fields
id
benchmarks/comm-rtx-5090-qwen3-8b-q4-tg-4kMetric
decode_tok_s
Value
185.9
Unit
tok/s TG batch-1 @4K ctx
Model
Qwen3-8B
Methodology
llama.cpp llama-bench, Q4_K_XL GGUF, CUDA, batch-1 decode (Hardware Corner measured table)
Measured by
third_party
Thermal state
unknown
notes
169.8 @8K, 111.9 @32K; PP 10,406 @4K. 14B 123.8 @4K. Qwen3-30B-A3B 226-234 @4K. 32B Q4 61.4 @4K / 50.9 @16K / 43.8 @32K.
Notes
169.8 @8K, 111.9 @32K; PP 10,406 @4K. 14B 123.8 @4K. Qwen3-30B-A3B 226-234 @4K. 32B Q4 61.4 @4K / 50.9 @16K / 43.8 @32K.