NVIDIA · Blackwell
Best AI Models for NVIDIA RTX PRO 5000 Blackwell (72 GB)
Runs 70B models well; 120B MoE only just fits.
72 GB GDDR71344 GB/s260 TFLOPS FP16300 WBlackwell 2025
| # | Model | Speed | llmrun Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 50.1 t/s | 50.1 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 50.1 t/s | 50.1 tok/s | Q4_K_M · 17.9 GB | |
| 3 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 42.8 t/s | 42.8 tok/s | Q4_K_M · 21.7 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 209 t/s | 209 tok/s | Q4_K_M · 16.5 GB | |
| 5 | Q4_K_M · 19.3 GB · 222 t/s | 222 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed | Coding Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 50.1 t/s | 50.1 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 42.8 t/s | 42.8 tok/s | Q4_K_M · 21.7 GB | |
| 3 | Qwen3.6 35B A3BAlibaba Q4_K_M · 22.0 GB · 181 t/s | 181 tok/s | Q4_K_M · 22.0 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 209 t/s | 209 tok/s | Q4_K_M · 16.5 GB | |
| 5 | Q4_K_M · 19.3 GB · 222 t/s | 222 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed | Reasoning Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 50.1 t/s | 50.1 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 42.8 t/s | 42.8 tok/s | Q4_K_M · 21.7 GB | |
| 3 | Qwen3.5 27BAlibaba Q4_K_M · 17.9 GB · 50.1 t/s | 50.1 tok/s | Q4_K_M · 17.9 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 209 t/s | 209 tok/s | Q4_K_M · 16.5 GB | |
| 5 | Q4_K_M · 19.3 GB · 222 t/s | 222 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed | Math Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 50.1 t/s | 50.1 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 50.1 t/s | 50.1 tok/s | Q4_K_M · 17.9 GB | |
| 3 | Qwen3.6 35B A3BAlibaba Q4_K_M · 22.0 GB · 181 t/s | 181 tok/s | Q4_K_M · 22.0 GB | |
| 4 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 42.8 t/s | 42.8 tok/s | overall | Q4_K_M · 21.7 GB |
| 5 | Q4_K_M · 19.3 GB · 222 t/s | 222 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed ↓ | llmrun Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.5 4BAlibabaHow to run → Q4_K_M · 3.4 GB · 268 t/s | 268 tok/s | Q4_K_M · 3.4 GB | |
| 2 | GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 233 t/s | 233 tok/s | Q4_K_M · 13.1 GB | |
| 3 | Q4_K_M · 19.3 GB · 222 t/s | 222 tok/s | Q4_K_M · 19.3 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 209 t/s | 209 tok/s | Q4_K_M · 16.5 GB | |
| 5 | Qwen3 30B A3B Thinking 2507Alibaba Q4_K_M · 19.4 GB · 200 t/s | 200 tok/s | Q4_K_M · 19.4 GB |
| # | Model | Speed | Vision Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 50.1 t/s | 50.1 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 50.1 t/s | 50.1 tok/s | Q4_K_M · 17.9 GB | |
| 3 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 42.8 t/s | 42.8 tok/s | Q4_K_M · 21.7 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 209 t/s | 209 tok/s | Q4_K_M · 16.5 GB | |
| 5 | MiniCPM V 4.7 35B A3BOpenBMB Q4_K_M · 22.1 GB · 200 t/s | 200 tok/s | Q4_K_M · 22.1 GB |
See all 379 models on the RTX PRO 5000 Blackwell →Best quant that fits · speeds estimated on this card
The RTX PRO 5000 Blackwell runs 70B models well. Blackwell generation, released 2025.
GenerationBlackwellreleased 2025
RTX PRO 5000 Blackwell strengths and weaknesses
Strengths
- Runs 120B MoE models: GPT OSS 120B at 168 tok/s in 70.8 GB
- Room for long context: Llama 3.3 70B Instruct at Q4 uses 51.3 GB, leaving 20.7 GB free
- Reads long prompts fast: A 4K-token prompt on Qwen3.8 27B starts answering in 1.9 s
- FP8 and NVFP4 run natively: Blackwell takes FP8 and NVFP4 checkpoints in vLLM as published
Output speed depends on memory bandwidth (1344 GB/s).
- 9BQwen3.5 9B · Q4_K_M · 6.6 GB137tok/s
- 14BQwen3 14B · Q4_K_M · 11.9 GB92.0tok/s
- 27–31BQwen3.8 27B · Q4_K_M · 17.9 GB50.1tok/s
- 35B MoEQwen3.6 35B A3B · Q4_K_M · 22.0 GB181tok/s
- 70BLlama 3.3 70B Instruct · Q4_K_M · 51.3 GB18.8tok/s
- 168tok/s
Estimated speeds: Q4_K_M, 4K context, one user. How we estimate
All 1253 models on the RTX PRO 5000 Blackwell
Featured onlyBest quant that fits, ordered by llmrun Score. Fit grades how well it runs here, S best. How ranking works
| Model | Quant | VRAM | Speed | Context | Score | Fit |
|---|---|---|---|---|---|---|
Olmo 3 7B ThinkChat | Q4_K_M | 8.4 GB | 151.9 t/s | 66K | S97, Runs great | |
CodeLlama 34B Instruct HFChatCode | Q4_K_M | 23.8 GB | 41.7 t/s | 16K | A76, Runs well | |
| Q4_K_M | 0.3 GB | 2912.0 t/s | 2K | S100, Runs great | ||
Nemotron Cascade 8BChatReasoning | Q4_K_M | 7.5 GB | 161.8 t/s | 33K | S97, Runs great | |
| Q4_K_M | 6.2 GB | 162.1 t/s | 8K | S97, Runs great | ||
Llama 160MChat | Q4_K_M | 0.5 GB | 1858.7 t/s | 2K | S100, Runs great | |
| Q4_K_M | 24.1 GB | 39.2 t/s | 524K | A75, Runs well | ||
| Q4_K_M | 7.5 GB | 122.4 t/s | 4K | S95, Runs great | ||
Codegen 350M MonoChatCode | FP16 | 0.8 GB | 1134.5 t/s | 2K | S100, Runs great | |
CodeLlama 70B Instruct HFChatCode | Q4_K_M | 43.0 GB | 20.6 t/s | 4K | B60, Decent | |
Qwen2 VL 2B InstructVision | Q4_K_M | 2.1 GB | 520.0 t/s | 33K | S100, Runs great | |
MiMo 7B BaseChat | Q4_K_M | 7.4 GB | 164.8 t/s | 33K | S97, Runs great | |
Open Llama 7BChat | Q4_K_M | 4.6 GB | 189.1 t/s | 2K | S99, Runs great | |
CodeLlama 7B Instruct HFChatCode | Q4_K_M | 12.9 GB | 161.2 t/s | 16K | S97, Runs great | |
Sqlcoder 7B 2ChatCode | Q4_K_M | 12.9 GB | 161.2 t/s | 16K | S97, Runs great |
Specs
- Architecture
- Blackwell
- VRAM
- 72 GB GDDR7
- Memory bandwidth
- 1344.0 GB/s
- FP16 (dense)
- 260.00 TFLOPS
- CUDA cores
- 14,080
- Tensor cores
- 440
- Compute capability
- 12.0
- TDP
- 300 W
- Released
- 2025-12-18
- MSRP
- $4,500
RTX PRO 5000 Blackwell FAQ
What is the largest model the RTX PRO 5000 Blackwell can run?
NVIDIA Nemotron 3 Super 120B A12B BF16 (123.6B) is the largest at 4-bit or better: at Q4_K_S it uses 70.0 GB of 72 GB, and runs at about 75.3 tok/s.
Can the RTX PRO 5000 Blackwell run Qwen3.8 27B?
Yes. Qwen3.8 27B fits at Q4_K_M, using 17.9 GB of 72 GB, and it is the best model that runs well on this card.
How many tokens per second does the RTX PRO 5000 Blackwell give?
About 50.1 tok/s on Qwen3.8 27B at Q4_K_M, with the first token after 1.9 s on a 4K-token prompt.
What power supply does the RTX PRO 5000 Blackwell need?
Plan a 600 W PSU (2× the 300 W TDP).
Is 72 GB enough for coding and agents?
Yes for coding: Qwen3.8 27B, the best of the 11 coding models that fit, runs at about 50 tok/s. For agents, not yet: no agent-scored model fits in less than 96 GB.