NVIDIA GeForce RTX 3090 Ti vs NVIDIA RTX PRO 6000 Blackwell
In short
The NVIDIA RTX PRO 6000 Blackwell has more memory for models: 96 GB against 24 GB. At 64k context the NVIDIA RTX PRO 6000 Blackwell runs 40 of the models we track fully in memory and the NVIDIA GeForce RTX 3090 Ti 22; only the NVIDIA RTX PRO 6000 Blackwell holds Qwen 3.8 Flash Next. On the models both run fully in memory, the NVIDIA RTX PRO 6000 Blackwell is about 1.7× as fast as the NVIDIA GeForce RTX 3090 Ti (estimated, median across models). In Germany the NVIDIA GeForce RTX 3090 Ti costs 2.500 € at Amazon and the NVIDIA RTX PRO 6000 Blackwell 19.076 € at eBay: the NVIDIA GeForce RTX 3090 Ti is 16.576 € cheaper.
Updated . Speeds are estimates; how they are worked out.
Specifications
| Spec | NVIDIA GeForce RTX 3090 Ti | NVIDIA RTX PRO 6000 Blackwell |
|---|---|---|
| GPU | GeForce RTX 3090 Ti | RTX PRO 6000 Blackwell |
| Memory | 24 GB GDDR6X | 96 GB GDDR7 |
| Memory bus | 384-bit | 512-bit |
| Memory bandwidth | 1,008 GB/s | 1,792 GB/s |
| FP16 compute | 80 TFLOPS | 504 TFLOPS |
| Software | CUDA | CUDA |
| Board power | 450 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 Ti | NVIDIA RTX PRO 6000 Blackwell |
|---|---|---|
| Qwen 3.8 Flash Next | won't run | 21tok/sUD-Q3_K_XL |
| Qwen 3.8 27B | 30tok/sUD-Q4_K_M | 33tok/sQ8_0 |
| MiniMax M2.7 | won't run | 33tok/sUD-Q2_K_XL |
| Qwen 3.5 122B-A10B | offload | 73tok/sQ5_K_M |
| Qwen 3.6 27B | 26tok/sQ5_K_M | 34tok/sQ8_0 |
| Gemma 4 31B | offload | 29tok/sQ8_0 |
| Qwen3 Coder Next | offload | 115tok/sQ8_0 |
| Nemotron 3 Super 120B-A12B | won't run | 77tok/sUD-Q4_K_M |
| gpt-oss 120B | won't run | 120tok/sMXFP4 |
| Mistral Small 4 119B | offload | 90tok/sUD-Q6_K |
| Qwen 3.6 35B-A3B | 115tok/sUD-Q4_K_M | 149tok/sQ8_0 |
| Qwen 3.5 27B | 30tok/sQ4_K_M | 34tok/sQ8_0 |
| Gemma 4 26B-A4B | 78tok/sUD-Q5_K_M | 117tok/sQ8_0 |
| GLM 4.5 Air | won't run | 26tok/sQ5_K_M |
| Qwen 3.5 35B-A3B | 116tok/sQ4_K_M | 149tok/sQ8_0 |
| GLM 4.7 Flash | 57tok/sQ4_K_M | 76tok/sQ8_0 |
| Devstral Small 2 24B | offload | 31tok/sQ8_0 |
| Qwen3 Next 80B-A3B | offload | 139tok/sQ8_0 |
| Nemotron 3.5 Lightning 30B-A3B | offload | 177tok/sUD-Q6_K_XL |
| gpt-oss 20B | 93tok/sMXFP4 | 165tok/sMXFP4 |
| Kimi Linear 48B-A3B | offload | 177tok/sQ8_0 |
| Qwen3 Coder 30B-A3B | offload | 69tok/sQ8_0 |
| Qwen3 32B | offload | 21tok/sQ8_0 |
| Gemma 4 12B | 44tok/sQ8_0 | 79tok/sQ8_0 |
| Qwen 3.5 9B | 53tok/sQ8_0 | 95tok/sQ8_0 |
| Gemma 3 27B | 28tok/sQ4_K_M | 32tok/sQ8_0 |
| Mistral Small 3.2 24B | offload | 31tok/sQ8_0 |
| DeepSeek R1 Distill Qwen 32B | offload | 21tok/sQ8_0 |
| Llama 3.3 70B | won't run | 12tok/sQ8_0 |
| Ministral 3 14B | 29tok/sQ6_K | 44tok/sQ8_0 |
| Qwen3 14B | 27tok/sQ6_K | 42tok/sQ8_0 |
| DeepSeek R1 0528 Qwen3 8B | 34tok/sQ8_0 | 61tok/sQ8_0 |
| Qwen3 8B | 34tok/sQ8_0 | 61tok/sQ8_0 |
| Phi-4 14B | 26tok/sQ5_K_M | 38tok/sQ8_0 |
| Qwen 3.5 4B | 94tok/sQ8_0 | 166tok/sQ8_0 |
| Ministral 3 8B | 34tok/sQ8_0 | 61tok/sQ8_0 |
| Granite 4.2 8B | 31tok/sQ8_0 | 55tok/sQ8_0 |
| Gemma 4 E4B | 62tok/sQ8_0 | 110tok/sQ8_0 |
| LFM2.5 8B-A1B | 88tok/sQ8_0 | 156tok/sQ8_0 |
| Llama 3.1 8B | 41tok/sQ6_K | 73tok/sQ6_K |
| DeepSeek V4 Flash | won't run | offload |
| GLM 5.3 Flash | won't run | offload |
Questions
- Which is better for local AI, the NVIDIA GeForce RTX 3090 Ti 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 Ti to spend less.
- Which can run bigger models, the NVIDIA GeForce RTX 3090 Ti or the NVIDIA RTX PRO 6000 Blackwell?
- At 64k context the NVIDIA RTX PRO 6000 Blackwell runs 40 of the models we track fully in memory and the NVIDIA GeForce RTX 3090 Ti 22; only the NVIDIA RTX PRO 6000 Blackwell holds Qwen 3.8 Flash Next.
- Which is faster for local AI, the NVIDIA GeForce RTX 3090 Ti or the NVIDIA RTX PRO 6000 Blackwell?
- On the models both run fully in memory, the NVIDIA RTX PRO 6000 Blackwell is about 1.7× as fast as the NVIDIA GeForce RTX 3090 Ti (estimated, median across models).
- What is the price difference between the NVIDIA GeForce RTX 3090 Ti and the NVIDIA RTX PRO 6000 Blackwell?
- In Germany the NVIDIA GeForce RTX 3090 Ti costs 2.500 € at Amazon and the NVIDIA RTX PRO 6000 Blackwell 19.076 € at eBay: the NVIDIA GeForce RTX 3090 Ti is 16.576 € cheaper.
More comparisons
- Mac Studio M3 Ultra 96GB vs NVIDIA RTX PRO 6000 Blackwell
- AMD Radeon AI PRO R9700 vs NVIDIA GeForce RTX 3090 Ti
- NVIDIA GeForce RTX 3090 vs NVIDIA GeForce RTX 3090 Ti
- NVIDIA GeForce RTX 3090 Ti vs NVIDIA GeForce RTX 4090
- NVIDIA GeForce RTX 3090 Ti vs NVIDIA GeForce RTX 5090
- NVIDIA GeForce RTX 3090 Ti vs AMD Radeon RX 7900 XTX
- NVIDIA GeForce RTX 5090 vs NVIDIA RTX PRO 6000 Blackwell