NVIDIA / discrete

NVIDIA RTX PRO 6000 Blackwell Workstation Edition

96 GB decides what fits. 1792 GB/s decides how fast it runs once it does.

Open in the calculator →Best models for 96 GB, on every card that size →

Find one: Amazon ↗ · eBay (new and used) ↗Store links may pay us a commission. They never decide which card is suggested — the memory arithmetic does.

Quick answer

The NVIDIA RTX PRO 6000 Blackwell Workstation Edition fits 245 of 320 sized models entirely in its 96 GB; the largest widely used one is Ling-3.0-flash (78.2 GB at Q4_K_M). Memory decides what fits; its 1792 GB/s of bandwidth decides how fast it answers.

Largest popular model that fitsLing-3.0-flash127B params · Q4_K_M · needs 78.2 GB~398 tok/s Faster than you readOpen in the calculator →
Best fast pickMixtral-8x22B-Instruct-v0.1141B params · Q4_K_M · needs 87.7 GB~48 tok/s Faster than you readOpen in the calculator →
Most downloaded that fitsQwen3.8-27B28B params · Q4_K_M · needs 18.6 GB~82 tok/s Faster than you readOpen in the calculator →

At 8,192 tokens of context with the whole model in device memory. Speeds are estimates from memory bandwidth, not benchmarks run on this card.

NVIDIA RTX PRO 6000 Blackwell Workstation Edition · 96 GB · 1792 GB/s · Q4_K_M where published · 8,192 tokensSpecificationpublished dataDecode speedestimate

245 of 320 fit entirely

245 of the 320 models the engine can size fit entirely in device memory at Q4_K_M where it is published, otherwise the nearest published format, and 8,192 tokens. 2 run with some layers on system memory (32 GB assumed), and 73 do not run at all.

77%of the models the engine can size fit entirely

96GB of device memory

1792GB/s memory bandwidth

2run with system memory

73do not run at 8,192 tokens

Featured models · Q4_K_M at 8,192 tokens, including offload

Model	Needs	Verdict	Decode	Calculator
Qwen3.5-2B2.3B parameters	2.32 GB	fits	~988 tok/s	Open →
NVIDIA-Nemotron-3-Nano-4B-BF164.0B parameters	3.46 GB	fits	~458 tok/s	Open →
Qwen3-VL-8B-Instruct8.8B parameters	7.04 GB	fits	~207 tok/s	Open →
Qwen3.5-9B9.7B parameters	7.05 GB	fits	~251 tok/s	Open →
gemma-4-12B-it12B parameters	9.54 GB	fits	~149 tok/s	Open →
gpt-oss-20b21B parameters	13.8 GB	fits	~487 tok/s	Open →
Qwen3.8-27B28B parameters	18.6 GB	fits	~82 tok/s	Open →
Qwen3.6-27B28B parameters	18.6 GB	fits	~82 tok/s	Open →
Kimi-Linear-48B-A3B-Instruct49B parameters	30.4 GB	fits	~652 tok/s	Open →
Qwen3.8-Flash-Next180B parameters	111.6 GB	7 of 48 layers on system RAM (16.1 GB)	~95 tok/s with offload	Open →
DeepSeek-V4-Flash-Vision-Exp305B parameters	188.1 GB	needs 95.8 GB of system RAM; 32 GB assumed	does not run	Open →
GLM-5.3-Flash321B parameters	198.3 GB	needs 105.2 GB of system RAM; 32 GB assumed	does not run	Open →

These examples are selected from prominent labs using the catalogue’s latest Hugging Face 30-day downloads and repository-creation freshness signal, with newer releases guaranteed a place. Offloaded rows assume 32 GB of system RAM, and a speed is only shown for a row that runs. Fitting in memory is not the same as loading: whether the runtime and version you have supports each architecture and format on this machine has not been tested here. Each “Open” link carries the same model, format, context and RAM into the calculator.

Fits entirely in NVIDIA RTX PRO 6000 Blackwell Workstation Edition memory — 245 of 320 sized models

Every model that fitswhole model resident · offload off

The most demanding model that fits is Mixtral-8x22B-Instruct-v0.1 at Q4_K_M: 87.7 GB of the 96 GB, leaving 8.29 GB spare.

Fully resident at 8,192 tokens (or the model’s own maximum, where that is shorter), offload off, at 1792 GB/s. Each row is a run of the engine for this configuration; the rows start with current, prominent releases and “fits” is memory, not a tested runtime. Older or less prominent models remain available through this search and “Show all”.

Find a model in this list

Showing 40 of 245 models that fit.

Model	Format	Needs	Spare	Decode	Downloads / 30d	Calculator
Qwen3.8-27BQwen · 28B params	Q4_K_M	18.6 GB	77.4 GB	~82 tok/s	6.9M	Open →
gemma-4-26B-A4B-itGoogle · 26B params	Q4_K_M	16.8 GB	79.2 GB	~458 tok/s	13M	Open →
gemma-4-31B-itGoogle · 31B params	Q4_K_M	22.2 GB	73.8 GB	~59 tok/s	9.9M	Open →
Qwen3.5-9BQwen · 9.7B params	Q4_K_M	7.05 GB	89.0 GB	~251 tok/s	9M	Open →
Qwen3.5-4BQwen · 4.7B params	Q4_K_M	3.98 GB	92.0 GB	~451 tok/s	7.8M	Open →
Qwen3.6-35B-A3BQwen · 36B params	Q4_K_M	23.1 GB	72.9 GB	~657 tok/s	3.3M	Open →
gemma-4-12B-itGoogle · 12B params	Q4_K_M	9.54 GB	86.5 GB	~149 tok/s	1.9M	Open →
Qwen3.6-27BQwen · 28B params	Q4_K_M	18.6 GB	77.4 GB	~82 tok/s	2.5M	Open →
Qwen3.5-2BQwen · 2.3B params	Q4_K_M	2.32 GB	93.7 GB	~988 tok/s	4.9M	Open →
gemma-4-E4B-itGoogle · 8.0B params	Q4_K_M	5.86 GB	90.1 GB	~403 tok/s	4.4M	Open →
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16NVIDIA · 32B params	Q4_K_M	20.3 GB	75.7 GB	~62 tok/s	530K	Open →
NVIDIA-Nemotron-3-Nano-4B-BF16NVIDIA · 4.0B params	Q4_K_M	3.46 GB	92.5 GB	~458 tok/s	3.4M	Open →
gemma-4-E2B-itGoogle · 5.1B params	Q4_K_M	4.02 GB	92.0 GB	~1064 tok/s	3M	Open →
Qwen3.5-0.8BQwen · 873M params	Q4_K_M	1.46 GB	94.5 GB	~2164 tok/s	2.6M	Open →
Qwen3-VL-8B-InstructQwen · 8.8B params	Q4_K_M	7.04 GB	89.0 GB	~207 tok/s	14.6M	Open →
Qwen3.5-27BQwen · 28B params	Q4_K_M	18.6 GB	77.4 GB	~82 tok/s	1.9M	Open →
North-Micro-Vision-InstructCohereLabs · 2.5B params	Q4_K_M	2.99 GB	93.0 GB	~650 tok/s	180.7K	Open →
Qwen3.5-35B-A3BQwen · 36B params	Q4_K_M	23.1 GB	72.9 GB	~657 tok/s	1.6M	Open →
NVIDIA-Nemotron-3-Super-120B-A12B-BF16NVIDIA · 124B params	Q4_K_M	76.9 GB	19.1 GB	~16 tok/sAbout reading pace	1.2M	Open →
granite-4.2-8bIBM · 8.8B params	Q4_K_M	7.49 GB	88.5 GB	~189 tok/s	128.4K	Open →
GLM-4.7-Flashzai-org · 31B params	Q4_K_M	20.4 GB	75.6 GB	~62 tok/s	1.8M	Open →
granite-4.1-3bIBM · 3.4B params	Q4_K_M	3.72 GB	92.3 GB	~440 tok/s	521.8K	Open →
LFM2.5-2.6BLiquidAI · 2.7B params	Q4_K_M	2.61 GB	93.4 GB	~706 tok/s	108.1K	Open →
Qwen3-VL-4B-InstructQwen · 4.4B params	Q4_K_M	4.72 GB	91.3 GB	~329 tok/s	3.5M	Open →
granite-4.1-30bIBM · 29B params	Q4_K_M	20.4 GB	75.6 GB	~62 tok/s	301.5K	Open →
Qwen3.5-122B-A10BQwen · 125B params	Q4_K_M	77.9 GB	18.1 GB	~241 tok/s	512.6K	Open →
Qwen3-VL-2B-InstructQwen · 2.1B params	Q4_K_M	3.05 GB	93.0 GB	~596 tok/s	2.8M	Open →
granite-4.2-3bIBM · 3.7B params	Q4_K_M	3.72 GB	92.3 GB	~440 tok/s	50.6K	Open →
LFM2.5-230MLiquidAI · 230M params	Q4_K_M	1.05 GB	94.9 GB	~4918 tok/s	88.6K	Open →
Qwen3-0.6BQwen · 752M params	Q4_K_M	2.22 GB	93.8 GB	~915 tok/s	29.7M	Open →
Qwen3-Coder-NextQwen · 80B params	Q4_K_M	50.0 GB	46.0 GB	~437 tok/s	596.3K	Open →
gpt-oss-20bOpenAI · 21B params	Q4_K_M	13.8 GB	82.2 GB	~487 tok/s	6.6M	Open →
granite-4.2-30bIBM · 29B params	Q4_K_M	20.7 GB	75.3 GB	~62 tok/s	35.8K	Open →
granite-4.1-8bIBM · 8.8B params	Q4_K_M	7.49 GB	88.5 GB	~189 tok/s	179.3K	Open →
NVIDIA-Nemotron-3-Nano-30B-A3B-BF16NVIDIA · 32B params	Q4_K_M	20.3 GB	75.7 GB	~62 tok/s	875.2K	Open →
gpt-oss-120bOpenAI · 117B params	Q4_K_M	71.9 GB	24.1 GB	~343 tok/s	4.5M	Open →
LFM2.5-VL-3BLiquidAI · 3.1B params	Q4_K_M	2.85 GB	93.1 GB	~706 tok/s	26.7K	Open →
MiMo-V2.6-Distill-Qwen-9BNEWXiaomiMiMo · 9.4B params	Q4_K_M	6.90 GB	89.1 GB	~251 tok/s	14.6K	Open →
Qwen3-4B-Instruct-2507Qwen · 4.0B params	Q4_K_M	4.72 GB	91.3 GB	~329 tok/s	3.7M	Open →
Ling-3.0-tinyinclusionAI · 7.9B params	Q4_K_M	5.85 GB	90.1 GB	~1421 tok/s	17.7K	Open →

Sized at the format most people actually download, not at FP16. The catalogue ordered by downloads →

Rent it instead

Rent an RTX PRO 6000 by the hour

What the NVIDIA RTX PRO 6000 Blackwell Workstation Edition costs on Vast.ai and RunPod right now, per machine, from one card to eight. Every price is read from the provider's own API.

Prices · 18:30 UTC, 22 Sept

1× RTX PRO 6000 · 96 GB

$1.14per hour

Vast.ai · Marketplace · 98.8% reliableRunPod $1.69/h Community CloudRent on Vast.ai

2× RTX PRO 6000 · 192 GB

$2.27per hour

Vast.ai · Marketplace · 98.8% reliableRent on Vast.ai

8× RTX PRO 6000 · 768 GB

$16.54per hour

Vast.ai · Marketplace · 98.4% reliableRent on Vast.ai

Buy one or rent one?

Your price, your hours, your electricity. Everything else is arithmetic.

Price you would pay ($)

Resale after 3 years ($)

Hours of use per day4 h

$ per kWh

Watts

Rent $/h

Watts start at the NVIDIA RTX PRO 6000 Blackwell Workstation Edition’s published board power (600 W), an upper bound: decoding rarely holds a card at its limit. Rent starts at the cheapest live price (Vast.ai). The electricity price is a placeholder — put yours in.

At 4 h a day, renting costs $1,664.40 a year.

That is $138.70 a month, and nothing when the machine is stopped.
Owning would cost $219.00 a year in electricity, on top of the price.
Every $1,000 of purchase price takes 0.7 years of this use to earn back.

Enter the price you would pay to see the exact break-even.

On-demand prices for the whole machine. Vast hosts below 98% measured reliability are left out; RunPod’s Community Cloud is vetted third-party hosts and its Secure Cloud is data-centre capacity. Every rentable card compared.

Referral links Vast.ai, RunPod and Novita pay us a share of what you spend if you sign up through these buttons. It costs you nothing, and it never decides an order or a recommendation: both are computed from the live price and the speed, and options that pay us nothing are listed and recommended on the same terms. How we rank

Find the NVIDIA RTX PRO 6000 Blackwell Workstation Edition: Amazon ↗ · eBay (new and used) ↗Store links may pay us a commission. They never decide which card is suggested — the memory arithmetic does.

Under the hood

The specification behind every figure

What the manufacturer publishes for this device, and the pages it was read from.

Manufacturer specification

Memory scope	dedicated
Capacity	96 GB
Published options	96 GB
Bandwidth	1792 GB/s
Memory type	GDDR7 ECC
Bus width	Not published
FP32 peak	125 TFLOPS
Dense matrix peak	Not published without sparsity
Max Power Consumption	600 W

Source ledger

NVIDIA RTX PRO 6000 Blackwell Workstation Edition ↗

Retrieved 2026-08-31

Caveats

The official page does not publish memory bus width; it remains blank rather than being derived.
No dense matrix throughput is catalogued for this device: NVIDIA publishes only "AI TOPS: 4000", footnoted "Theoretical FP4 TOPS using sparsity". Time to first token is withheld rather than priced against a sparsity figure.

Published capacity is a hardware ceiling, not guaranteed free runtime memory. The calculator shows the runtime reserve separately rather than folding it into a single number.

Run the diagnostic on the NVIDIA RTX PRO 6000 Blackwell Workstation Edition →