Calculate the TPS of the RTX PRO 4000 Blackwell on local AI models
Every model in our catalogue assessed against this card at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from this card's memory bandwidth and the size of each model once compressed.
Calculated for this card
Largest model it holds
Mixtral 8x7B
46.7B · Q3_K_M · 59.5 tok/s
Fastest model
Gemma 3 QAT 1B
285 tok/s · 1B
What AI models can a RTX PRO 4000 Blackwell run?
Set the inputs, read the answer
More context means more memory for the conversation cache. Speed is for a fresh conversation and does not change with this setting.
Hides models that would only fit by being compressed below this point.
502 models match
Calculating| Quantisation | Fit | ||||||
|---|---|---|---|---|---|---|---|
|
285
tok/s
242–342 |
Gemma 3 1B | 1B | Mar 2025 | 1.8 GB | 33k tokens | Q8_0 | Comfortable |
|
285
tok/s
242–342 |
Gemma 3 QAT 1B | 1B | Apr 2025 | 1.8 GB | 33k tokens | Q8_0 | Comfortable |
|
285
tok/s
171–455 · low confidence |
HGRN 1B (WT 103) ≈ | 1B | Nov 2023 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
285
tok/s
171–455 · low confidence |
LLama 3..2 Typhoon 2 1B ≈ | 1B | Dec 2024 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
285
tok/s
171–455 · low confidence |
OLMo-1B ≈ | 1B | Feb 2024 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
285
tok/s
171–455 · low confidence |
Pythia-1b ≈ | 1B | Apr 2023 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
264
tok/s
158–422 · low confidence |
OpenELM-1.1B ≈ | 1.1B | May 2024 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
259
tok/s
155–414 · low confidence |
DeciCoder-1B ≈ | 1.1B | Aug 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
259
tok/s
155–414 · low confidence |
SantaCoder ≈ | 1.1B | Jan 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
259
tok/s
155–414 · low confidence |
TinyLlama-1.1B (1T token checkpoint) ≈ | 1.1B | Oct 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
259
tok/s
155–414 · low confidence |
TinyLlama-1.1B (3T token checkpoint) ≈ | 1.1B | Oct 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
237
tok/s
142–379 · low confidence |
EXAONE 4.0 (1.2B) ≈ | 1.2B | Jul 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
237
tok/s
142–379 · low confidence |
MinerU2.5 ≈ | 1.2B | Sep 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
237
tok/s
142–379 · low confidence |
Pleias 1.0 1.2B ≈ | 1.2B | Dec 2024 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
237
tok/s
142–379 · low confidence |
Pleias-RAG-1B ≈ | 1.2B | Apr 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
231
tok/s
197–278 |
Llama 3.2 1B | 1.2B | Sep 2024 | 2.2 GB | 131k tokens | Q8_0 | Comfortable |
|
228
tok/s
137–365 · low confidence |
MiniCPM-1.2B ≈ | 1.2B | Jun 2024 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
219
tok/s
131–350 · low confidence |
DeepSeek Coder 1.3B ≈ | 1.3B | Jan 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
219
tok/s
131–350 · low confidence |
DeepSeek-VL-1.3B ≈ | 1.3B | Mar 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
219
tok/s
131–350 · low confidence |
DigiRL ≈ | 1.3B | Jun 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
219
tok/s
131–350 · low confidence |
GLA Transformer 1.3B ≈ | 1.3B | Aug 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
219
tok/s
131–350 · low confidence |
Janus 1.3B ≈ | 1.3B | Oct 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
219
tok/s
131–350 · low confidence |
Kosmos-2.5 ≈ | 1.3B | Aug 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
219
tok/s
131–350 · low confidence |
Otter ≈ | 1.3B | May 2023 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
219
tok/s
131–350 · low confidence |
Phi-1 ≈ | 1.3B | Oct 2023 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.
On record
RTX PRO 4000 Blackwell full specification
Everything on record for this board, ordered by how much it bears on running a language model rather than by how a spec sheet would list it. Memory comes first because it decides the outcome; the rest is context.
Memory
The two specifications that decide what this card can run and how quickly. Capacity sets which models fit; bandwidth sets how many tokens per second they produce once they do.
- Memory size
- 24 GB
- Memory bandwidth
- 672 GB/s
- Memory type
- GDDR7
- Memory bus width
- 192 bit
- Memory clock
- 1.75 GHz
The chip
Which processor is on the board and how it was manufactured. A smaller process size generally means more performance for the same power.
- Graphics processor
- GB203
- Architecture
- Blackwell 2.0
- Generation
- Blackwell PRO W(x000)
- Foundry
- TSMC
- Process size
- 5 nm
- Transistors
- 45.6 billion
- Transistor density
- 120,600 K/mm²
- Die size
- 378 mm²
- Released
- 18 March 2025
Clock speeds
How fast the processor runs. Worth far less here than on a gaming benchmark: generating text is limited by memory bandwidth, so a higher clock barely moves the result.
- Base clock
- 1.23 GHz
- Boost clock
- 2.06 GHz
Processing units
What the chip contains. These drive graphics performance and matter mainly for processing a long prompt rather than for producing the answer.
- Shading units
- 8,960
- Texture mapping units
- 280
- Render output units
- 96
- Streaming multiprocessors
- 70
- Tensor cores
- 280
- Ray tracing cores
- 70
- L1 cache
- 128 KB
- L2 cache
- 48 MB
Theoretical performance
Peak arithmetic rates published for the board. These are ceilings that no real workload reaches, and generating text reaches a small fraction of them because it is limited by memory rather than arithmetic.
- Half precision (FP16)
- 36.8 TFLOPS
- Single precision (FP32)
- 36.8 TFLOPS
- Double precision (FP64)
- 575.4 GFLOPS
- Pixel rate
- 197 GPixel/s
- Texture rate
- 575 GTexel/s
The board
What it takes to physically install and power the card — the practical constraints that decide whether it fits the machine you already own.
- Power draw (TDP)
- 140 W
- Suggested power supply
- 300 W
- Power connectors
- 1x 16-pin
- Bus interface
- PCIe 5.0 x16
- Slot width
- Single-slot
- Dimensions
- 241 mm × 20 mm
- Display outputs
- 4x DisplayPort 2.1b
Software support
Which graphics and compute interfaces the card supports. CUDA compute capability is the one that bears on inference: below 7.0 there are no tensor cores, and modern inference software falls back to slower code paths.
- CUDA compute capability
- 12.0
- DirectX
- 12.2
- OpenGL
- 4.6
- Vulkan
- 1.4
- OpenCL
- 3.0
- Shader model
- 6.8
Listings
Where to buy a RTX PRO 4000 Blackwell
No vendor is currently listing this card. Listings come from vendors who publish them here directly — browse the vendor directory to see who is selling what.
What the numbers mean
Memory: the specification that decides everything
Memory
24 GB
Bandwidth
672 GB/s
Largest model
Mixtral 8x7B
The RTX PRO 4000 Blackwell carries 24 GB of GDDR7, which covers the mid-sized models most people actually run — about 21.6 GB of it after the runtime and driver reserve their working space.
The memory bus moves 672 GB/s across a 192-bit bus. That is the number that governs generation speed — arithmetic per byte read is small enough that the bus, not the cores, is what everything waits on.
Bandwidth is clock times bus width, and this card clocks its memory at 1.75 GHz. Both halves matter, and neither is visible in a gaming benchmark.
Put together, the largest model that fits is Mixtral 8x7B at 46.7B, running Q3_K_M and producing around 59.5 tokens per second.
The chip and how it was built
The RTX PRO 4000 Blackwell is built on the GB203 graphics processor, using NVIDIA's Blackwell 2.0 architecture, as part of the Blackwell PRO W(x000) generation.
The chip is manufactured by TSMC, on a 5 nm process, with a die measuring 378 mm², holding 45.6 billion transistors. A smaller process generally means more performance for the same power, though for language models it matters far less than the memory subsystem.
It was released in March 2025, roughly 1 years ago. Inference software support tends to follow hardware by a year or two, so a card of this age generally has mature, well-optimised code paths available to it.
Compute throughput, and why it matters less than it looks
FP16
36.8 TFLOPS
FP64
575.4 GFLOPS
Tensor cores
280
On paper the RTX PRO 4000 Blackwell reaches 36.8 TFLOPS at half precision and 36.8 TFLOPS at single precision. These are peak figures no real workload sustains, and generating text reaches only a small fraction of them — decoding is limited by memory rather than arithmetic, which is why a card can look enormously powerful here and still produce tokens at an ordinary rate.
Double-precision throughput is 575.4 GFLOPS. It has no bearing on running a language model — no inference runtime uses it — but it separates datacentre parts from consumer ones, since the latter deliberately restrict it.
The card carries 280 tensor cores across 70 streaming multiprocessors. These accelerate the matrix arithmetic at the heart of a transformer, and they are what make prompt processing — reading a long document before answering — dramatically faster than it would otherwise be.
Clocks run from 1.23 GHz at base to 2.06 GHz boosted. Worth far less here than on a gaming benchmark: raising the clock speeds up the arithmetic, and the arithmetic is not what generation is waiting on.
Cache and processing units
The RTX PRO 4000 Blackwell has 128 KB of L1 cache, backed by 48 MB of L2. Cache absorbs a share of the memory traffic that would otherwise hit the main bus, which is the one place on this page where a number other than bandwidth quietly affects generation speed — a large L2 lets more of the working set stay close to the cores.
There are 8,960 shading units, 280 texture mapping units, and 96 render output units. These drive graphics workloads and contribute to prompt processing, but they sit idle for much of the time a model spends generating a reply.
Power, size and installation
Power draw
140 W
The RTX PRO 4000 Blackwell is rated at 140 W, with a 300 W power supply suggested for the whole system. Running a language model keeps a card busy in bursts rather than continuously — it draws hard while generating and idles between requests — so sustained draw over a working day is usually well below the rated figure.
The board occupies a single-slot, measuring 241 mm long, and needs 1x 16-pin. Worth checking against the case and power supply already in the machine, since the largest cards need considerably more of both than a typical desktop provides.
It connects over PCIe 5.0 x16. The interface governs how quickly a model is loaded from disk into the card, not how fast it runs once there, so a narrower link costs a few seconds at startup and nothing thereafter.
The extremes
The largest AI models a RTX PRO 4000 Blackwell can run
The biggest open-weight models that fit on this card, newest first. Each is shown at the best compression the card can hold.
The fastest AI models on a RTX PRO 4000 Blackwell
Where this card produces tokens quickest. Smaller models dominate here, because generating each token means reading the whole model out of memory once.
Step by step
How to work out the tokens per second of a RTX PRO 4000 Blackwell
You do not have to calculate anything by hand — the gputps.com calculator on this page has already worked it out for every model this card can hold. Reading off the answer takes six steps.
-
01
Start with the model, not the specification
The table lists 502 models this RTX PRO 4000 Blackwell can run. Search by name, or by size — typing 27b matches on the parameter count even when the name never states it.
-
02
Match the context to your work
Drag the slider to the conversation length you plan to work at. The cache grows with the conversation, and on 24 GB it is often what pushes a large model over the edge.
-
03
Set a minimum quality if you need one
By default the table picks the least-compressed copy that fits. Setting a floor removes models that only qualify through heavy compression.
-
04
Read the speed and the range
Speeds come with error bars for a reason. The best case here is 285 tok/s on Gemma 3 QAT 1B, and which inference software you use moves that by thirty to fifty per cent.
-
05
Read the fit verdict last
The fit column separates models that just fit from those with room to spare — worth checking against the card's 24 GB before settling on one.
-
06
Cross-check against other hardware
Each model page repeats this calculation for the whole catalogue. Worth a look before deciding: it shows what else runs the same model, and how the RTX PRO 4000 Blackwell compares.
Answers
RTX PRO 4000 Blackwell — common questions
Can a RTX PRO 4000 Blackwell run a model that does not fit in its memory?
Only partly. Layers beyond the 24 GB sit in system memory and run at a fraction of the speed, so a mostly-offloaded model is rarely worth using. Every figure here assumes it is fully resident on the card.
Would two RTX PRO 4000 Blackwell cards be twice as fast?
Capacity adds, throughput does not. Two of them give you 48 GB to work with rather than twice the tokens per second — every figure here is for a single RTX PRO 4000 Blackwell.
What AI models can a RTX PRO 4000 Blackwell run?
502 of the 679 open-weight language models we track fit on a RTX PRO 4000 Blackwell and can be run locally on it. The table on this page lists every one, with the memory it needs, the quantisation it runs at and an estimated generation speed.
What is the largest AI model a RTX PRO 4000 Blackwell can run?
The largest model in our catalogue that fits on a RTX PRO 4000 Blackwell is Mixtral 8x7B at 46.7B parameters, compressed to Q3_K_M. It generates roughly 59.5 tokens per second and needs about 21.0 GB of the card's memory.
How many tokens per second does a RTX PRO 4000 Blackwell produce?
It depends on the model. On a RTX PRO 4000 Blackwell the fastest model we track is Gemma 3 QAT 1B at about 285 tokens per second, while larger models run proportionally slower because each token requires reading the whole model out of memory once. Speeds are estimates for a single conversation at a time.
Can a RTX PRO 4000 Blackwell run a 7B model?
Yes. For example a RTX PRO 4000 Blackwell runs Multi-Token Prediction 7B at Q8_0, using about 7.9 GB of memory and generating around 42.5 tokens per second.
Can a RTX PRO 4000 Blackwell run a 13B model?
Yes. For example a RTX PRO 4000 Blackwell runs DeepSeekMoE-16B at Q8_0, using about 17.5 GB of memory and generating around 98.8 tokens per second.
Can a RTX PRO 4000 Blackwell run a 30B model?
Yes. For example a RTX PRO 4000 Blackwell runs Nemotron 3-Nano-30B-A3B at Q4_K_M, using about 18.1 GB of memory and generating around 116 tokens per second.
How much memory does a RTX PRO 4000 Blackwell have?
A RTX PRO 4000 Blackwell has 24 GB of GDDR7 memory. Around a tenth of that is reserved by the inference runtime and the driver, leaving roughly 21.6 GB available for a model and its conversation.
What is the memory bandwidth of a RTX PRO 4000 Blackwell?
The RTX PRO 4000 Blackwell has 672 GB/s of memory bandwidth, across a 192-bit memory bus. This is the single best predictor of how fast it generates text, because producing each token means reading the entire model out of memory once.
What type of memory does a RTX PRO 4000 Blackwell use?
It uses GDDR7 clocked at 1.75 GHz. HBM types are found on datacentre accelerators and carry far more bandwidth than the GDDR used on desktop cards, which is why they generate tokens considerably faster at the same capacity.
Who makes the RTX PRO 4000 Blackwell?
The RTX PRO 4000 Blackwell is a NVIDIA product, with the chip manufactured by TSMC, on a 5 nm process.
When was the RTX PRO 4000 Blackwell released?
The RTX PRO 4000 Blackwell was released in March 2025.
How much power does a RTX PRO 4000 Blackwell use?
The RTX PRO 4000 Blackwell has a rated board power of 140 W, and a 300 W system power supply is suggested. Generating text draws hard in bursts and idles between requests, so average consumption over a working session is normally well below the rated figure.
How much cache does a RTX PRO 4000 Blackwell have?
The RTX PRO 4000 Blackwell has 128 KB of L1 cache, and 48 MB of L2 cache. Cache absorbs part of the memory traffic that would otherwise reach the main bus, so a larger L2 gives a modest lift to generation speed beyond what bandwidth alone predicts.
What are the TFLOPS of a RTX PRO 4000 Blackwell?
The RTX PRO 4000 Blackwell is rated at 36.8 TFLOPS at half precision and 36.8 TFLOPS at single precision. These are peak arithmetic ceilings rather than achievable rates, and text generation reaches only a small fraction of them because it is limited by memory bandwidth instead.
How many tensor cores does a RTX PRO 4000 Blackwell have?
The RTX PRO 4000 Blackwell has 280 tensor cores across 70 streaming multiprocessors. They accelerate the matrix arithmetic a transformer is built from, which mainly speeds up processing a long prompt rather than producing the reply.
Does the RTX PRO 4000 Blackwell support CUDA?
Yes. The RTX PRO 4000 Blackwell reports CUDA compute capability 12.0. Capability 7.0 and above has tensor cores, which modern inference software uses; below that it falls back to slower code paths for quantised models.
What bus interface does the RTX PRO 4000 Blackwell use?
It uses PCIe 5.0 x16. This governs how fast a model is loaded onto the card rather than how fast it runs once loaded, so it costs a few seconds at startup and nothing during generation.
Is the RTX PRO 4000 Blackwell good for running local AI models?
Its memory comfortably covers the mid-sized models most people run locally and its bandwidth gives usable, if unspectacular, generation speeds. In total it runs 502 of the models we track. Whether that is enough depends entirely on which model you want — the table above answers that directly.
The other direction
Looking at it from the other side?
This page starts from the hardware. If you already know which model you want and need to know what it takes to run it, start from the model instead.