NVIDIA RTX 4090 48GB (China-mod)
No editorial image yet — generic vendor mark shown. Credentials in spec table below.
Third-party physical modification of a stock GIGABYTE / ASUS / MSI RTX 4090. Chinese specialty shops (and GPVLab in the US) rebuild the card to 48 GB. GPVLab documents its process: the AD102 core and the 12 stock GDDR6X modules move to a new clamshell PCB with 12 added modules, 24 × 2 GB in total, on the same 384-bit memory bus. GPVLab lists 900 GB/s bandwidth for both stock and upgraded cards; NVIDIA lists 1,008 GB/s for a stock 4090.
Affiliate disclosure: as an Amazon Associate and partner of other retailers, we earn from qualifying purchases. The verdict on this page is our editorial opinion; affiliate links never influence what we recommend.
Sub-scores sum to 763 / 1000. Headline = 763 × 0.70 (Estimated-confidence discount) = 534. This is an algorithmic performance-tier score — distinct from, and often lower than, the editorial “Our verdict” below, which weighs value and real-world fit (especially for hardware we haven’t measured yet). How scoring works →
Extrapolated from 1008 GB/s bandwidth — 121.0 tok/s estimated. No measured benchmarks yet.
Plain-English: Runs 70B with care — snappy enough for a coding agent; vision models supported.
Verdicts extrapolated from catalog VRAM + bandwidth + ecosystem flags. Hover any chip for the rationale. Want measured numbers? Submit your own run with runlocalai-bench --submit.
Overview
Third-party physical modification of a stock GIGABYTE / ASUS / MSI RTX 4090. Chinese specialty shops (and GPVLab in the US) rebuild the card to 48 GB. GPVLab documents its process: the AD102 core and the 12 stock GDDR6X modules move to a new clamshell PCB with 12 added modules, 24 × 2 GB in total, on the same 384-bit memory bus. GPVLab lists 900 GB/s bandwidth for both stock and upgraded cards; NVIDIA lists 1,008 GB/s for a stock 4090. The 48 GB capacity opens up workloads that the stock 4090 can't host: Llama 3.3 70B Q4 with comfortable context, dual-32B models in parallel, longer context windows on 32B-class models. **GamersNexus has covered the modification process** (youtube.com/watch?v=TcRGBeOENLg). Community sellers list on Taobao and Bilibili; GPVLab takes US/EU orders. Important caveats: not a vendor-blessed product. No NVIDIA warranty. GPVLab lists the same 450 W TGP as a stock 4090. Real-world benchmark data is community-reported and not yet in our editorial benchmark queue.
Some links above are affiliate links. We may earn a commission at no extra cost to you. How we make money.
Specs
| VRAM | 48 GB |
| Power draw (peak) | 450 W |
| Released | 2024 |
| Backends | CUDA Vulkan |
Models that fit
Open-weight models small enough to run on NVIDIA RTX 4090 48GB (China-mod) with usable context.
Frequently asked
What models can NVIDIA RTX 4090 48GB (China-mod) run?
Does NVIDIA RTX 4090 48GB (China-mod) support CUDA?
Where next?
Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify hardware specifications.