← AI Hashrate 中文

RTX 4090 24GB vs Titan Xp 12GB for Local LLMs

The RTX 4090 24GB has about 84% more memory bandwidth (1,008 vs 548 GB/s). Batch-1 decode tracks bandwidth closely, so a model that fits both cards should decode roughly 84% faster on the RTX 4090 24GB. The RTX 4090 24GB fits more of the catalog at Q4/8K — 35 vs 18 of 69 models (24 vs 12 GB VRAM). Largest model that fits the RTX 4090 24GB but not the Titan Xp 12GB: Qwen3.6-35B-A3B (36.0B). MSRP is $1,599 for the RTX 4090 24GB vs $1,200 for the Titan Xp 12GB. MSRP is a launch list price, not live retail — Amazon prices move; click through for the current price. In the Q4/8K head-to-head below (41 shared models), the RTX 4090 24GB posts the higher tok/s in 41 rows vs 0. Methodology.

Spec comparison

RTX 4090 24GBTitan Xp 12GB
VendorNVIDIANVIDIA
VRAM24 GB12 GB
Memory typeGDDR6XGDDR5X
Memory bandwidth1,008 GB/s548 GB/s
TDP450 W250 W
MSRP (list)$1,599$1,200
BuyGIGABYTE GeForce RTX 4090 Gaming OC 24GB (affiliate) · More listingsSearch Amazon (affiliate)

MSRP is the launch list price, not live retail. Buy links are Amazon affiliate links — the price after you click is the live price.

Q4 / 8K head-to-head — top 15 of 41 shared models

ModelParams (B)RTX 4090 24GB tok/sRTX 4090 24GB fitsTitan Xp 12GB tok/sTitan Xp 12GB fitsWinner
MiniCPM5-1B1.08593.9 est.Yes322.9 est.YesRTX 4090 24GB
Qwen3.5-2B2.0320.7 est.Yes174.4 est.YesRTX 4090 24GB
DeepSeek-R1-0528-Qwen3-8B-layer-mix-bpw-3.8-mlx2.3278.9 est.Yes151.6 est.YesRTX 4090 24GB
gpt-oss-20b21.5225.2 measuredYes216.2 est.NoRTX 4090 24GB
Qwen3.6-35B-A3B36.0213.8 est.Yes109.2 est.NoRTX 4090 24GB
Qwen3.5-35B-A3B34.7213.8 est.Yes117.2 est.NoRTX 4090 24GB
GLM-4.7-Flash31.2213.8 est.Yes135.3 est.NoRTX 4090 24GB
Qwen3-30B-A3B-Instruct-250730.5213.8 est.Yes141.0 est.NoRTX 4090 24GB
Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF1630.0213.8 est.Yes155.0 est.NoRTX 4090 24GB
Ornith-1.0-35B35.0213.8 est.Yes109.6 est.NoRTX 4090 24GB
qwen35b-a3b-fable-sft-abliterated34.7213.8 est.Yes111.3 est.NoRTX 4090 24GB
Llama-3.2-3B-Instruct3.2200.5 est.Yes109.0 est.YesRTX 4090 24GB
Mistral-7B-Instruct-v0.37.2183.3 measuredYes48.4 est.YesRTX 4090 24GB
Qwen3-Coder-Next79.7174.1 est.No23.6 est.NoRTX 4090 24GB
gemma-4-26B-A4B-it25.2168.8 est.Yes154.1 est.NoRTX 4090 24GB

est. = bandwidth-based estimate; measured = curated real benchmark that overrides the estimate. Fits = weights + KV(8K) + 1 GB overhead ≤ 95% of VRAM.

Fit coverage at Q4 / 8K

RTX 4090 24GBTitan Xp 12GB
Catalog models that fit35 of 6918 of 69
Largest exclusive fitQwen3.6-35B-A3B (36.0B)

RTX 4090 24GB details → · Titan Xp 12GB details →