NVIDIA
Operating mode
Use this to bias workload recommendations toward responsiveness, background autonomy, lighter serving, or multi-GPU scale-out.
Current mode
Balanced
Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.
The RTX PRO 6000 Blackwell Workstation Edition is the most powerful single workstation GPU ever built, featuring 96 GB of ECC GDDR7 at 1,792 GB/s bandwidth with 125 TFLOPS FP16 and 4,000 INT8 TOPS. Based on the GB202 die with 24,064 CUDA cores, it can run 70B models at FP16 on a single card and fits 100B+ models at Q4 — previously achievable only with multi-GPU data-center setups. Available from April 2025, it also supports Universal MIG for partitioning the GPU into multiple isolated inference instances.
Beyond LLMs
What AI tasks this GPU can handle — from text generation to image and video creation.
| Capability | Status | Representative Model |
|---|---|---|
| LLM Chat (7B) | Runs natively | Llama 3.1 8B Q4 |
| LLM Coding (30B) | Runs natively | Qwen 3 30B Q4 |
| LLM Large (70B) | Runs natively | Llama 3.1 70B Q4 |
| Image Gen (SDXL) | Runs natively | SDXL 1.0 FP16 |
| Image Gen (Flux) | Runs natively | Flux.1 Dev FP16 |
| Image Gen (SD 3.5) | Runs natively | SD 3.5 Large FP16 |
| Video Short (25f) | Runs natively | LTX Video 2B |
| Video Long (100f) | Tight fit | Wan Video 14B |
Architecture
Blackwell is NVIDIA's fifth-generation RTX architecture, built on TSMC's 4NP process. It introduces 5th-generation Tensor Cores with native FP4 precision support, enabling double the inference throughput per watt compared to Ada Lovelace's FP8 operations. Key innovations include the Neural Rendering Pipeline for AI-driven shading and the debut of GDDR7 memory in consumer GPUs.
AI Relevance
FP4 Tensor Cores deliver the highest tokens-per-watt efficiency in any consumer architecture. Native FP4 quantization means models can run at lower precision with minimal quality loss, effectively doubling the effective VRAM for model weights.
Conselho de compra
Excelente escolha para IA local
Roda 35 de 50 modelos principais bem — um ótimo coringa para inferência local.
96.0 GB
VRAM
$9,999
Preço sugerido
$104/GB
Custo por GB de VRAM
Melhores modelos para esta GPU
What will limit you first
This setup is broadly balanced for this model.
No major red flags
This recommendation has enough memory headroom and acceptable estimated speed for the selected workload.
Best upgrade itinerary
Unlocks 2 additional models that do not fit on the current setup.
Quer mais margem? NVIDIA H200 141GB (141.0 GB VRAM) é o próximo passo.
Chat
SQwen3-Coder-Next is viable for Chat, but is not the most specialized choice. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
Coding
SQwen3-Coder-Next is a specialized fit for Coding. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
Agentic Coding
SQwen3-Coder-Next is a specialized fit for Agentic Coding. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
Reasoning
SQwen3-Coder-Next matches Reasoning and keeps a practical fit profile. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
RAG
SQwen 3.5 27B matches RAG and keeps a practical fit profile. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
Qwen3-Coder-NextQwen 3.5 122B A10B
Mistral Small 4 119BQwen 2.5 VL 72BDevstral 2 123B InstructQwen 3.6 35B A3BQwen3-Coder 30B A3B InstructQwen 3.5 27BQwen3-VL 30B A3B InstructQwen 3.5 35B A3BQwen 3 32B
Command A 111B
GPT-OSS 120BMagistral Small 2507Devstral Small 2 24B InstructQwen 3.6 27BQwen 3 30B A3BNemotron 3 Nano 30BPixtral Large 124BLeanstral 119B A6BDevstral Small 1.1Qwen 3.5 9BQwen 3 14B
Gemma 4 31BNemotron Cascade 2 30B A3B
Phi-4-reasoning-plus 14BGPT-OSS 20BQwen 3 8B
EXAONE 4.0 32BQwen 3.5 4BMinistral 3 14BNemotron Nano 8BPhi-4 Mini Reasoning 4B
Jina Embeddings v3
BGE M3Qwen 3.5 397B A17B
Kimi K2.5Kimi K2.6
DeepSeek V4 ProDeepSeek V4 Flash
GLM-5.2GLM-5.1GLM-5DeepSeek V3.2Qwen 3 235B A22BQwen3-Coder 480B A35B InstructDeepSeek Coder V2 236BDeepSeek R1 671BDeepSeek V3.1 671BQuase ao alcance
Modelos de alta qualidade que precisam de um pouco mais de memória
Image & Video Generation
55 of 56 models can generate images or video on your RTX PRO 6000 Blackwell Workstation Edition 96GB
| Model | Max Resolution | Gen Time | Grade |
|---|---|---|---|
| SD TurboImage | 512×512 | 300ms | S |
| Stable Diffusion 1.5Image | 512×768 | 600ms | S |
| Realistic Vision v5.1Image | 512×768 | 600ms | S |
| DreamShaper 8Image | 512×768 | 600ms | S |
| LCM DreamShaper v7Image | 512×768 | 200ms | S |
| PixArt-SigmaImage | 1024×1024 | ~2.4s | S |
| FramePack I2VVideo | 1280×720 | ~4.4s/frame | S |
| SDXL TurboImage | 512×512 | 300ms | S |
| SDXL LightningImage | 1024×1024 | 900ms | S |
| Stable Diffusion XL 1.0Image | 1024×1024 | ~2.4s | S |
| Playground v2.5Image | 1024×1024 | ~3.6s | S |
| RealVisXL v5.0Image | 1024×1024 | ~2.7s | S |
| DreamShaper XLImage | 1024×1024 | ~2.7s | S |
| Juggernaut XL v9Image | 1024×1024 | ~2.7s | S |
| Animagine XL 3.1Image | 1024×1024 | ~2.7s | S |
| Pony Diffusion V6 XLImage | 1024×1024 | ~2.7s | S |
| Animagine XL 4.0Image | 1024×1024 | ~2.7s | S |
| Illustrious XLImage | 1024×1024 | ~2.7s | S |
| Wan Video 2.1 1.3BVideo | 480×832 | ~1.7s/frame | S |
| Stable Diffusion 3.5 MediumImage | 1024×1024 | ~4.2s | S |
| Flux.2 Klein 4BImage | 1024×1024 | 700ms | S |
| LTX Video 2BVideo | 1280×720 | ~2.1s/frame | S |
| KolorsImage | 1024×1024 | ~4.7s | S |
| Stable CascadeImage | 1024×1024 | ~5.9s | S |
| AuraFlow v0.3Image | 1536×1536 | ~10.7s | S |
| Stable Diffusion 3.5 LargeImage | 1024×1024 | ~13.1s | S |
| Stable Diffusion 3.5 Large TurboImage | 1024×1024 | ~2.4s | S |
| CogVideoX 2BVideo | 720×480 | ~2.1s/frame | S |
| HunyuanVideoVideo | 720×1280 | ~4.4s/frame | S |
| ChromaImage | 1024×1024 | ~2.4s | S |
| Z-Image TurboImage | 1536×1536 | ~2.4s | S |
| Flux.1 DevImage | 1024×1024 | ~10.7s | S |
| Flux.1 SchnellImage | 1024×1024 | ~2.1s | S |
| LTX Video 13BVideo | 1280×720 | ~4.4s/frame | S |
| Flux.1 Kontext DevImage | 1024×1024 | ~11.9s | S |
| AnimateDiff v1.5.3Video | 512×768 | ~1.1s/frame | S |
| Cosmos Diffusion 7BVideo | 1024×576 | ~3.4s/frame | S |
| CogVideoX 5BVideo | 720×480 | ~3s/frame | S |
| Wan2.2 TI2V 5BVideo | 832×480 | ~3s/frame | S |
| Flux.2 Klein 9BImage | 1024×1024 | ~1.2s | S |
| Flux.1 Fill DevImage | 1024×1024 | ~10.1s | S |
| Krea 2Image | 1024×1024 | ~3.2s | S |
| Sulphur 2Video | 1280×720 | ~3.8s/frame | S |
| Ideogram 4Image | 2048×2048 | ~2.9s | S |
| Mochi 1 PreviewVideo | 848×480 | ~3.9s/frame | S |
| HunyuanVideo 1.5Video | 720×1280 | ~3.6s/frame | S |
| Helios 14BVideo | 1280×720 | ~4.5s/frame | S |
| SkyReels V2 14BVideo | 1280×720 | ~4.5s/frame | S |
| Wan Video 2.1 14BVideo | 720×1280 | ~4.5s/frame | S |
| Wan Video 2.2 14BVideo | 720×1280 | ~4.5s/frame | S |
| Qwen ImageImage | 1024×1024 | ~4s | S |
| Qwen Image EditImage | 1024×1024 | ~4s | S |
| LTX-2 22BVideo | 1920×1088 | ~5.4s/frame | S |
| Flux.2 DevImage | 1024×1024 | ~1m 52s | S |
| MAGI-1Video | 1280×720 | ~5.6s/frame | S |
| HunyuanImage 3.0Image | 256×256 | ~7s | F |
Image models estimated at 1024×1024 (28 steps, FP16). Video models estimated at 768×512 (25 frames, 30 steps, FP16). Actual performance varies with runtime and system load.
Upgrade paths
See what you unlock with more powerful hardware
Opções de upgrade
NVIDIA H200 141GBPróximo passoUnlocks 2 additional models that do not fit on the current setup.
Lifts average decode speed across fitting models by about 51%.
~$30,000 MSRP
NVIDIA B200 180GBUpgrade NVIDIAUnlocks 8 additional models that do not fit on the current setup.
Lifts average decode speed across fitting models by about 89%.
~$30,000 MSRP
Unlocks 12 additional models that do not fit on the current setup.
Lifts average decode speed across fitting models by about 55%.
~$20,000 MSRP
Unlocks 13 additional models that do not fit on the current setup.
Lifts average decode speed across fitting models by about 75%.
~$8,000 MSRP
RTX PRO 6000 Blackwell Workstation Edition 96GB (96 GB VRAM) can run these top models: Qwen3-Coder-Next (score: 95/100), Qwen 3.5 122B A10B (score: 95/100), Mistral Small 4 119B (score: 93/100). See the full compatibility list above.
RTX PRO 6000 Blackwell Workstation Edition 96GB has 96 GB of VRAM available for AI model inference. This determines which models and quantization levels you can run locally.
Yes, RTX PRO 6000 Blackwell Workstation Edition 96GB is excellent for running LLMs locally with top compatibility scores above 80/100.
For coding on RTX PRO 6000 Blackwell Workstation Edition 96GB, we recommend Qwen3-Coder-Next. It achieves 79.4 tokens per second with 217K context window. Qwen3-Coder-Next is a specialized fit for Coding. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
There are 4 upgrade path(s) from RTX PRO 6000 Blackwell Workstation Edition 96GB: NVIDIA H200 141GB, NVIDIA B200 180GB. Upgrading would unlock larger models and faster inference speeds.
Yes, RTX PRO 6000 Blackwell Workstation Edition 96GB with 96 GB of usable memory can run Flux.1 Dev at FP16 natively. Flux is a 12B parameter diffusion transformer that produces high-quality images. You can also run the Schnell variant for faster generation.
RTX PRO 6000 Blackwell Workstation Edition 96GB (96 GB VRAM) can handle various AI generation tasks beyond LLMs. For image generation, SDXL and Stable Diffusion 3.5 run well. Flux.1 Dev also runs natively for state-of-the-art image quality. For video, LTX Video 2.3 can generate short clips. Check the AI Capability Matrix above for detailed compatibility.
RTX PRO 6000 Blackwell Workstation Edition 96GB is excellent for AI image generation. With 96 GB of usable memory, it runs all major diffusion models including Flux.1, SDXL, and Stable Diffusion 3.5 at full precision. You can generate high-resolution images quickly and even handle video generation models.
Yes, RTX PRO 6000 Blackwell Workstation Edition 96GB with 96 GB of usable memory can run Qwen 3.5 27B at Q8 (near-lossless, ~28.9 GB) or even FP16 (~55.4 GB) depending on your context needs. This setup provides an excellent experience with this model. Use Ollama or vLLM for best results.
With 96 GB VRAM on RTX PRO 6000 Blackwell Workstation Edition 96GB, use Q8_0 for most models — it is near-lossless and you have the memory for it. For 70B+ models, Q6_K offers excellent quality. Reserve Q4_K_M for 100B+ models or when you need maximum context length.
RTX PRO 6000 Blackwell Workstation Edition 96GB already has strong memory bandwidth, so the next limit is often memory capacity and context headroom rather than raw decode speed. For local LLMs, fit first and bandwidth second is the right mental model.
Compare with similar
