NVIDIA RTX PRO 5000 Blackwell (48 GB) for local LLMs
Written by Jakub Rusinowski · Last updated
With 48 GB of GDDR7 at 1,344 GB/s, the RTX PRO 5000 Blackwell (48 GB) runs 115 catalogued models at Q4_K_M with 8K context. The largest that fits is Llama 3.3 70B Instruct (~45.7 GB), and the top pick is Apertus 70B at 16–33 tok/s.
48 GB of GDDR7 with ECC, 1,344 GB/s, 300 W, dual slot. The same card is also sold with 72 GB at the same bandwidth; that is a separate record.
Models that run on the RTX PRO 5000 Blackwell (48 GB)
Q4_K_M, 8K context, 48 GB usable. Ranked by quality and speed.
| Model | Memory · marker = 48 GB | VRAM | Speed |
|---|---|---|---|
| Apertus 70B Apertus | 43.1 GB | 16–33 tok/s | |
| Cosmos 3 Super Cosmos 3 | 39.4 GB | 31–64 tok/s | |
| Qwen 3.6 35B-A3B Qwen 3.6 | 21.9 GB | 119–247 tok/s | |
| Nex-N2.5 mini Nex-N2.5 | 21.9 GB | 119–247 tok/s | |
| Nex-N2 mini Nex-N2 | 21.9 GB | 119–247 tok/s | |
| Qwen 3.5 35B-A3B Qwen 3.5 | 21.9 GB | 119–247 tok/s | |
| Laguna XS 2.1 33B-A3B Poolside Laguna XS 2.1 | 20.7 GB | 119–248 tok/s | |
| Qwen 3 32B Qwen 3 | 22.8 GB | 30–63 tok/s | |
| Aya Expanse 32B Aya Expanse | 21.6 GB | 32–66 tok/s | |
| EXAONE 3.5 32B EXAONE 3.5 | 22.3 GB | 31–64 tok/s | |
| Cogito v1 32B Cogito v1 | 22.3 GB | 31–64 tok/s | |
| Granite 4.0 Small-H 32B-A9B IBM Granite 4.0 | 20.1 GB | 75–157 tok/s |
Buy it or rent the same memory
Buy the card, or rent a GPU with the same memory by the hour to try models first.
As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.
Speed vs other GPUs
Llama 3.1 8B, Q4_K_M. Estimated ranges. How this is calculated
Specifications
Specs last updated 2026-10-07.
- Memory
- 48 GB GDDR7
- Memory bandwidth
- 1,344 GB/s
- Memory bus
- 384-bit
- Architecture
- Blackwell GB202
- Series
- RTX PRO Blackwell
- Board power
- 300 W
- Release year
- 2025
- Compute backends
- CUDA, VULKAN
- Usable for models
- 48 GB
Similar GPUs
Frequently asked questions
Can the NVIDIA RTX PRO 5000 Blackwell (48 GB) run local LLMs?
Yes. With 48 GB (48 GB usable by a model) it runs 115 of the catalogued models at Q4_K_M with 8K context; the largest is Llama 3.3 70B Instruct, needing about 45.7 GB.
How fast is the NVIDIA RTX PRO 5000 Blackwell (48 GB) for AI inference?
It is estimated to run Llama 3.1 8B at 94–194 tok/s at Q4_K_M. Llama 3.3 70B is estimated at 16–33 tok/s. These are modelled estimates from memory bandwidth, not measurements; the methodology page shows the formula.
What LLMs can I run on 48 GB?
Among the best that fit: Apertus 70B, Cosmos 3 Super, Qwen 3.6 35B-A3B, Nex-N2.5 mini, Nex-N2 mini.