Skip to content
llm-speed

RTX 5090 vs RTX 3090

Signed, community-submitted decode tok/s, prefill, and TTFT for RTX 5090 vs RTX 3090 — every number links to the run it came from.

Verdict

RTX 5090 and RTX 3090 both have submitted runs, but no single model has been measured on both yet — so there is no apples-to-apples row. Each side's measured decode tok/s is listed separately below (peaks of 247 and 189 tok/s). Submit a shared workload to turn this into a direct comparison.

hardware
RTX 5090
NVIDIA Ada/Blackwell · 32 GB VRAM
View RTX 5090 page →
hardware
RTX 3090
NVIDIA Ampere · 24 GB VRAM
View RTX 3090 page →

RTX 5090 benchmarks — no shared model with RTX 3090 yet

Modeldecode tok/sWorkloadRun
Qwen2.5-7B-Instruct246.7tok/schat-shortr_3yn-4321hp-
Qwen2.5-Coder-7B-Instruct244.9tok/schat-shortr_mln72x5zbis
Llama-3.1-8B-Instruct232.2tok/schat-shortr_kfrkg-vn376
Qwen3.6-35B-A3B-Q4_K_M.gguf224.0tok/schat-shortr_a56-wxl21lk
Qwen2.5-Coder-14B-Instruct136.0tok/schat-shortr_p_f63tcgans
Qwen3.6-27B-Q4_K_M.gguf73.31tok/schat-shortr_wax_x2ryqhk
Qwen2.5-Coder-32B-Instruct70.96tok/sconcurrent-decoder_v983y0y3r2u
Codestral-22B-v0.169.95tok/schat-shortr_sqr8liqh4ii
Qwen3-32B69.45tok/schat-shortr_phvxm9dcak0
gemma-2-9b-it69.45tok/schat-shortr_1_xl4zb5-xj
gpt-oss-20b69.42tok/schat-shortr_r9h57uts9lr
gemma-4-31B-it-Q4_K_M.gguf67.37tok/schat-shortr_moawmhigq5d

RTX 3090 benchmarks — no shared model with RTX 5090 yet

Modeldecode tok/sWorkloadRun
deepseek-coder-v2189.5tok/schat-shortr_o2-1w665rtq
qwen2.5-coder69.21tok/schat-shortr_6ahy-dq0f_0

See also: RTX 5090 benchmarks · RTX 3090 benchmarks · All hardware · All models · Methodology