NVIDIA GeForce RTX 3090 vs NVIDIA RTX PRO 6000 Blackwell
For running models locally, priced in Canada. At 8k context the first runs 28 models fully in memory, the second 41.
In short
The NVIDIA RTX PRO 6000 Blackwell has more memory for models: 96 GB against 24 GB. At 8k context the NVIDIA RTX PRO 6000 Blackwell runs 41 of the models we track fully in memory and the NVIDIA GeForce RTX 3090 28; only the NVIDIA RTX PRO 6000 Blackwell holds DeepSeek V4 Flash. On the models both run fully in memory, the NVIDIA RTX PRO 6000 Blackwell is about 1.8× as fast as the NVIDIA GeForce RTX 3090 (estimated, median across models). In Canada the NVIDIA GeForce RTX 3090 costs $2,800 at Amazon and the NVIDIA RTX PRO 6000 Blackwell $3,133 at Amazon: the NVIDIA GeForce RTX 3090 is $333 cheaper.
Updated . Speeds are estimates; how they are worked out.
Specifications
| Spec | NVIDIA GeForce RTX 3090 | NVIDIA RTX PRO 6000 Blackwell |
|---|---|---|
| GPU | GeForce RTX 3090 | RTX PRO 6000 Blackwell |
| Memory | 24 GB GDDR6X | 96 GB GDDR7 |
| Memory bus | 384-bit | 512-bit |
| Memory bandwidth | 936 GB/s | 1,792 GB/s |
| FP16 compute | 71 TFLOPS | 504 TFLOPS |
| Software | CUDA | CUDA |
| Board power | 350 W | 600 W |
| PCIe generation | Gen 4 | Gen 5 |
| PCIe lanes | x16 | x16 |
What each runs
Estimated decode speed with the most precise quant that fits each one est.
| Model | NVIDIA GeForce RTX 3090 | NVIDIA RTX PRO 6000 Blackwell |
|---|---|---|
| DeepSeek V4 Flash | won't run | 89tok/sUD-Q2_K_XL |
| Qwen 3.8 Flash Next | won't run | 21tok/sUD-Q3_K_XL |
| Qwen 3.8 27B | 26tok/sUD-Q6_K | 37tok/sQ8_0 |
| MiniMax M2.7 | won't run | 117tok/sUD-Q2_K_XL |
| Qwen 3.5 122B-A10B | offload | 79tok/sQ6_K |
| Qwen 3.6 27B | 25tok/sQ6_K | 38tok/sQ8_0 |
| Gemma 4 31B | 25tok/sQ5_K_M | 33tok/sQ8_0 |
| Qwen3 Coder Next | offload | 151tok/sQ8_0 |
| Nemotron 3 Super 120B-A12B | won't run | 81tok/sUD-Q4_K_M |
| gpt-oss 120B | won't run | 195tok/sMXFP4 |
| Mistral Small 4 119B | offload | 110tok/sUD-Q6_K |
| Qwen 3.6 35B-A3B | 168tok/sUD-Q4_K_M | 201tok/sQ8_0 |
| Qwen 3.5 27B | 25tok/sQ6_K | 38tok/sQ8_0 |
| Gemma 4 26B-A4B | 88tok/sUD-Q6_K | 147tok/sQ8_0 |
| GLM 4.5 Air | offload | 40tok/sQ6_K |
| Qwen 3.5 35B-A3B | 168tok/sQ4_K_M | 201tok/sQ8_0 |
| GLM 4.7 Flash | 87tok/sQ5_K_M | 117tok/sQ8_0 |
| Devstral Small 2 24B | 28tok/sQ6_K | 42tok/sQ8_0 |
| Qwen3 Next 80B-A3B | offload | 196tok/sQ8_0 |
| Nemotron 3.5 Lightning 30B-A3B | offload | 195tok/sUD-Q6_K_XL |
| gpt-oss 20B | 132tok/sMXFP4 | 253tok/sMXFP4 |
| Qwen3 Coder 30B-A3B | 118tok/sQ5_K_M | 167tok/sQ8_0 |
| Kimi Linear 48B-A3B | offload | 202tok/sQ8_0 |
| Qwen3 32B | 27tok/sQ4_K_M | 30tok/sQ8_0 |
| Gemma 4 12B | 44tok/sQ8_0 | 85tok/sQ8_0 |
| Qwen 3.5 9B | 59tok/sQ8_0 | 113tok/sQ8_0 |
| Mistral Small 3.2 24B | 28tok/sQ6_K | 42tok/sQ8_0 |
| DeepSeek R1 Distill Qwen 32B | 26tok/sQ4_K_M | 30tok/sQ8_0 |
| Gemma 3 27B | 25tok/sQ6_K | 37tok/sQ8_0 |
| Llama 3.3 70B | offload | 14tok/sQ8_0 |
| Ministral 3 14B | 37tok/sQ8_0 | 71tok/sQ8_0 |
| Qwen3 14B | 34tok/sQ8_0 | 65tok/sQ8_0 |
| DeepSeek R1 0528 Qwen3 8B | 59tok/sQ8_0 | 112tok/sQ8_0 |
| Qwen3 8B | 59tok/sQ8_0 | 112tok/sQ8_0 |
| Phi-4 14B | 34tok/sQ8_0 | 64tok/sQ8_0 |
| Qwen 3.5 4B | 121tok/sQ8_0 | 231tok/sQ8_0 |
| Ministral 3 8B | 57tok/sQ8_0 | 109tok/sQ8_0 |
| Granite 4.2 8B | 54tok/sQ8_0 | 104tok/sQ8_0 |
| Gemma 4 E4B | 69tok/sQ8_0 | 131tok/sQ8_0 |
| LFM2.5 8B-A1B | 237tok/sQ8_0 | 455tok/sQ8_0 |
| Llama 3.1 8B | 76tok/sQ6_K | 145tok/sQ6_K |
| GLM 5.3 Flash | won't run | offload |
Questions
- Which is better for local AI, the NVIDIA GeForce RTX 3090 or the NVIDIA RTX PRO 6000 Blackwell?
- Choose the NVIDIA RTX PRO 6000 Blackwell to run bigger models. Choose the NVIDIA RTX PRO 6000 Blackwell for speed. Choose the NVIDIA GeForce RTX 3090 to spend less.
- Which can run bigger models, the NVIDIA GeForce RTX 3090 or the NVIDIA RTX PRO 6000 Blackwell?
- At 8k context the NVIDIA RTX PRO 6000 Blackwell runs 41 of the models we track fully in memory and the NVIDIA GeForce RTX 3090 28; only the NVIDIA RTX PRO 6000 Blackwell holds DeepSeek V4 Flash.
- Which is faster for local AI, the NVIDIA GeForce RTX 3090 or the NVIDIA RTX PRO 6000 Blackwell?
- On the models both run fully in memory, the NVIDIA RTX PRO 6000 Blackwell is about 1.8× as fast as the NVIDIA GeForce RTX 3090 (estimated, median across models).
- What is the price difference between the NVIDIA GeForce RTX 3090 and the NVIDIA RTX PRO 6000 Blackwell?
- In Canada the NVIDIA GeForce RTX 3090 costs $2,800 at Amazon and the NVIDIA RTX PRO 6000 Blackwell $3,133 at Amazon: the NVIDIA GeForce RTX 3090 is $333 cheaper.
More comparisons
- Mac Studio M3 Ultra 96GB vs NVIDIA RTX PRO 6000 Blackwell
- AMD Radeon AI PRO R9700 vs NVIDIA GeForce RTX 3090
- AMD Radeon AI PRO R9700 vs NVIDIA RTX PRO 6000 Blackwell
- NVIDIA GeForce RTX 3090 vs NVIDIA GeForce RTX 3090 Ti
- NVIDIA GeForce RTX 3090 vs NVIDIA GeForce RTX 4070 Ti Super
- NVIDIA GeForce RTX 3090 vs NVIDIA GeForce RTX 4080 Super
- NVIDIA GeForce RTX 3090 vs NVIDIA GeForce RTX 5070 Ti
- NVIDIA GeForce RTX 3090 vs NVIDIA GeForce RTX 5080
- NVIDIA GeForce RTX 3090 vs NVIDIA GeForce RTX 5090
- NVIDIA GeForce RTX 3090 vs AMD Radeon RX 7900 XTX