NVIDIA RTX 4090 vs NVIDIA RTX Pro 4000 Blackwell for local LLMs
35 reports on the NVIDIA RTX 4090 and 3 on the NVIDIA RTX Pro 4000 Blackwell
Which is faster for local LLMs?
Nobody has reported the same model at the same quant on both cards yet, so there's no measured head-to-head. Memory bandwidth, which caps how fast a card can write, is 1.5x higher on the NVIDIA RTX 4090 (1,008 GB/s on the NVIDIA RTX 4090, 672 on the NVIDIA RTX Pro 4000 Blackwell). The estimates below compare them on the models people run most.
- Memory
- 24GB
- Memory bandwidth
- 1,008 GB/s
- FP16 compute
- 330 TFLOPS
- Reports
- 35
- Memory
- 24GB
- Memory bandwidth
- 672 GB/s
- FP16 compute
- 161.3 TFLOPS
- Reports
- 3
Estimated by the calculator
The quant the calculator recommends for each card and its estimated speed at a 32,768-token context with 32 GB of system RAM, from memory bandwidth and the model's shape.
| Model | NVIDIA RTX 4090 | NVIDIA RTX Pro 4000 Blackwell |
|---|---|---|
| Qwen3.8 27B | 36.5 t/sQ4_K_M | 24.3 t/sQ4_K_M |
| Qwen3.8 125B · 6B active | 19.6 t/sQ2_K, experts in RAM | 18.8 t/sQ2_K, experts in RAM |
| DeepSeek V4 Flash 284B · 13B active | doesn't fit at these settings | doesn't fit at these settings |
| Qwen3.6 35B · 3B active | 129.3 t/sQ4_K_S | 86.2 t/sQ4_K_S |
| Qwen3.6 27B | 36.5 t/sQ4_K_M | 24.3 t/sQ4_K_M |
| Gemma 4 26B · 4B active | 82.3 t/sQ5_K_S | 54.9 t/sQ5_K_S |
Change the context, RAM or model in the calculator: NVIDIA RTX 4090 · NVIDIA RTX Pro 4000 Blackwell
Every report on each card, with its source: NVIDIA RTX 4090 · NVIDIA RTX Pro 4000 Blackwell