RTX PRO 5000 48GB Blackwell
NVIDIA RTX PRO 5000 48GB Blackwell: 48 GB of memory at 1344 GB/s. Below: the open models it can run, estimated at 48 GB with an 8K context. 48 GB at 1,344 GB/s: enough for a 70B dense model at 4-bit on one card.
Last updated: 2026-10-16
Specifications
What it can run
Best-quality quantization that fits with a 8K context, and the estimated generation speed. Estimates, not benchmarks. See the methodology.
| Model | Params | Best quant | Memory | tok/s (est.) |
|---|---|---|---|---|
| Llama 3.3 70B Instruct | 70B | Q4_K_M | 44.5 GiB | 21 |
| Qwen3.5-35B-A3B | 35B (3B) | Q8_0 | 37.0 GiB | 186 |
| Qwen3.6-35B-A3B | 35B (3B) | Q8_0 | 37.0 GiB | 186 |
| Qwen3-32B | 32.8B | Q8_0 | 36.6 GiB | 26 |
| DeepSeek-R1-Distill-Qwen-32B | 32.76B | Q8_0 | 36.5 GiB | 26 |
| Gemma 4 31B IT | 30.7B | Q8_0 | 33.5 GiB | 27 |
| Qwen3-30B-A3B (2507) | 30.5B (3.3B) | Q8_0 | 32.9 GiB | 175 |
| Qwen3-Coder-30B-A3B-Instruct | 30.5B (3.3B) | Q8_0 | 32.9 GiB | 175 |
| GLM-4.7-Flash (30B-A3B) | 30B (3B) | Q8_0 | 32.1 GiB | 186 |
| Qwen3.5-27B | 27B | Q8_0 | 29.1 GiB | 31 |
| Qwen3.6-27B | 27B | Q8_0 | 29.1 GiB | 31 |
| Qwen3.8-27B | 27B | Q8_0 | 29.1 GiB | 31 |
| Gemma 3 27B IT | 27B | Q8_0 | 29.6 GiB | 31 |
| Gemma 4 26B A4B IT | 25.2B (3.8B) | Q8_0 | 27.0 GiB | 159 |
| Mistral Small 3.2 24B Instruct (2506) | 24B | Q8_0 | 26.7 GiB | 34 |
| Magistral Small 1.2 (2509) | 24B | Q8_0 | 26.7 GiB | 34 |
| Devstral Small 2 24B Instruct (2512) | 24B | Q8_0 | 26.7 GiB | 34 |
| gpt-oss-20b | 21B (3.6B) | FP16 | 41.8 GiB | 104 |
| Phi-4 (14B) | 14B | FP16 | 29.4 GiB | 31 |
| Gemma 3 12B IT | 12B | FP16 | 24.8 GiB | 36 |
| Gemma 4 12B Unified IT | 11.95B | FP16 | 24.2 GiB | 36 |
| Qwen3.5-9B | 9B | FP16 | 18.4 GiB | 47 |
| Qwen3-8B | 8.2B | FP16 | 17.7 GiB | 51 |
| DeepSeek-R1-0528-Qwen3-8B | 8.19B | FP16 | 17.6 GiB | 52 |
| Llama 3.1 8B Instruct | 8B | FP16 | 17.1 GiB | 53 |
| Gemma 4 E4B IT | 8B | FP16 | 16.4 GiB | 53 |
| Llama 3.2 3B Instruct | 3.21B | FP16 | 7.7 GiB | 113 |
| Kimi K2.6 | 1000B (32B) | Doesn't fit | – | – |
| DeepSeek-R1-0528 (671B) | 671B (37B) | Doesn't fit | – | – |
| DeepSeek-V4-Flash | 284B (13B) | Doesn't fit | – | – |
| Qwen3.8-Flash-Next | 180B (6B) | Doesn't fit | – | – |
| Qwen3.5-122B-A10B | 122B (10B) | Doesn't fit | – | – |
| Mistral Small 4 119B (2603) | 119B (6.5B) | Doesn't fit | – | – |
| gpt-oss-120b | 117B (5.1B) | Doesn't fit | – | – |
| Llama 4 Scout 17B-16E Instruct | 109B (17B) | Doesn't fit | – | – |
| GLM-4.5-Air | 106B (12B) | Doesn't fit | – | – |
| Qwen3-Coder-Next (80B-A3B) | 80B (3B) | Doesn't fit | – | – |
Prices
Latest price at each shop, as listed on the day we checked.
| Region | Shop | Listing | Price | Per GB | Checked |
|---|---|---|---|---|---|
| Taiwan | 原價屋 CoolPC | Leadtek RTX PRO 5000 Blackwell 48GBLine on the shop's public price list (no per-product page); list stamped 2026/10/15 9:31 | NT$325,000 | NT$6,771/GB | 2026-10-15 |
- Taiwan value: about 139.8 tok/s on Llama 3.1 8B Instruct (Q4_K_M, 8K), or 0.4 tok/s per NT$1,000.
A price history chart appears once we have checked this device on two or more days.
All prices →Frequently asked questions
Which LLMs can the RTX PRO 5000 48GB Blackwell run?
With 48 GB and an 8K context, the largest models that fit include: Llama 3.3 70B Instruct, Qwen3.5-35B-A3B, Qwen3.6-35B-A3B, Qwen3-32B, DeepSeek-R1-Distill-Qwen-32B. See the full table below.
How fast is the RTX PRO 5000 48GB Blackwell for local LLMs?
Generation speed is mostly set by memory bandwidth (1344 GB/s). For Llama 3.1 8B Instruct at FP16, expect roughly 53 tokens/s; bigger models are slower.
Can the RTX PRO 5000 48GB Blackwell run 70B models?
Yes. Llama 3.3 70B Instruct fits on the RTX PRO 5000 48GB Blackwell (48 GB) at Q4_K_M with an 8K context, using about 44.5 GiB, at an estimated 21 tokens/s.