Calculate the TPS of the Data Center GPU Max Subsystem 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
Solar Open2 250B
250.3B · Q3_K_M · 52.9 tok/s
Fastest model
Gemma 3 QAT 1B
884 tok/s · 1B
What AI models can a Data Center GPU Max Subsystem 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.
624 models match
Calculating| Quantisation | Fit | ||||||
|---|---|---|---|---|---|---|---|
|
884
tok/s
530–1,414 · low confidence |
Gemma 3 1B | 1B | Mar 2025 | 1.8 GB | 33k tokens | Q8_0 | Comfortable |
|
884
tok/s
530–1,414 · low confidence |
Gemma 3 QAT 1B | 1B | Apr 2025 | 1.8 GB | 33k tokens | Q8_0 | Comfortable |
|
884
tok/s
530–1,414 · low confidence |
HGRN 1B (WT 103) ≈ | 1B | Nov 2023 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
884
tok/s
530–1,414 · low confidence |
LLama 3..2 Typhoon 2 1B ≈ | 1B | Dec 2024 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
884
tok/s
530–1,414 · low confidence |
OLMo-1B ≈ | 1B | Feb 2024 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
884
tok/s
530–1,414 · low confidence |
Pythia-1b ≈ | 1B | Apr 2023 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
818
tok/s
491–1,309 · low confidence |
OpenELM-1.1B ≈ | 1.1B | May 2024 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
803
tok/s
482–1,285 · low confidence |
DeciCoder-1B ≈ | 1.1B | Aug 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
803
tok/s
482–1,285 · low confidence |
SantaCoder ≈ | 1.1B | Jan 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
803
tok/s
482–1,285 · low confidence |
TinyLlama-1.1B (1T token checkpoint) ≈ | 1.1B | Oct 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
803
tok/s
482–1,285 · low confidence |
TinyLlama-1.1B (3T token checkpoint) ≈ | 1.1B | Oct 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
736
tok/s
442–1,178 · low confidence |
EXAONE 4.0 (1.2B) ≈ | 1.2B | Jul 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
736
tok/s
442–1,178 · low confidence |
MinerU2.5 ≈ | 1.2B | Sep 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
736
tok/s
442–1,178 · low confidence |
Pleias 1.0 1.2B ≈ | 1.2B | Dec 2024 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
736
tok/s
442–1,178 · low confidence |
Pleias-RAG-1B ≈ | 1.2B | Apr 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
718
tok/s
431–1,150 · low confidence |
Llama 3.2 1B | 1.2B | Sep 2024 | 2.2 GB | 131k tokens | Q8_0 | Comfortable |
|
708
tok/s
425–1,133 · low confidence |
MiniCPM-1.2B ≈ | 1.2B | Jun 2024 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
680
tok/s
408–1,088 · low confidence |
DeepSeek Coder 1.3B ≈ | 1.3B | Jan 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
680
tok/s
408–1,088 · low confidence |
DeepSeek-VL-1.3B ≈ | 1.3B | Mar 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
680
tok/s
408–1,088 · low confidence |
DigiRL ≈ | 1.3B | Jun 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
680
tok/s
408–1,088 · low confidence |
GLA Transformer 1.3B ≈ | 1.3B | Aug 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
680
tok/s
408–1,088 · low confidence |
Janus 1.3B ≈ | 1.3B | Oct 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
680
tok/s
408–1,088 · low confidence |
Kosmos-2.5 ≈ | 1.3B | Aug 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
680
tok/s
408–1,088 · low confidence |
Otter ≈ | 1.3B | May 2023 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
680
tok/s
408–1,088 · 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
Data Center GPU Max Subsystem 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
- 128 GB
- Memory bandwidth
- 3,210 GB/s
- Memory type
- HBM2e
- Memory bus width
- 8,192 bit
- Memory clock
- 1.57 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
- Ponte Vecchio
- Architecture
- Generation 12.5
- Generation
- Data Center GPU(Ponte Vecchio)
- Foundry
- Intel
- Process size
- 10 nm
- Transistors
- 100 billion
- Transistor density
- 78,100 K/mm²
- Die size
- 1,280 mm²
- Released
- 10 January 2023
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
- 900 MHz
- Boost clock
- 1.6 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
- 16,384
- Texture mapping units
- 1,024
- Ray tracing cores
- 128
- L1 cache
- 64 KB
- L2 cache
- 408 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)
- 52.4 TFLOPS
- Single precision (FP32)
- 52.4 TFLOPS
- Double precision (FP64)
- 52.4 TFLOPS
- Texture rate
- 1,638 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)
- 2,400 W
- Suggested power supply
- 2,800 W
- Power connectors
- 1x 16-pin
- Bus interface
- PCIe 5.0 x16
- Slot width
- Dual-slot
- Dimensions
- 267 mm
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.
- DirectX
- 12.1
- OpenGL
- 4.6
- OpenCL
- 3.0
- Shader model
- 6.6
Listings
Where to buy a Data Center GPU Max Subsystem
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
128 GB
Bandwidth
3,210 GB/s
Largest model
Solar Open2 250B
With 128 GB of HBM2e, the Data Center GPU Max Subsystem is in the class of hardware that holds the largest open-weight models without splitting them across machines. Roughly 115.2 GB of that is reachable by an inference runtime once the driver takes its share.
Its 3,210 GB/s across a 8,192-bit bus is at the top of what exists. Since each token means reading the whole model out of memory once, that translates almost directly into generation speed — this card is bandwidth-rich enough that model size stops being the limiting factor long before the bus does.
That comes from a 1.57 GHz memory clock across the bus width above. Widening the bus and raising the clock are the two levers a manufacturer has, which is why a card with unremarkable cores can still generate quickly.
The practical ceiling is Solar Open2 250B at 250.3B, held at Q3_K_M and running at roughly 52.9 tokens per second.
The chip and how it was built
The Data Center GPU Max Subsystem is built on the Ponte Vecchio graphics processor, using Intel's Generation 12.5 architecture, as part of the Data Center GPU(Ponte Vecchio) generation.
The chip is manufactured by Intel, on a 10 nm process, with a die measuring 1,280 mm², holding 100 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 January 2023, roughly 3 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
52.4 TFLOPS
FP64
52.4 TFLOPS
On paper the Data Center GPU Max Subsystem reaches 52.4 TFLOPS at half precision and 52.4 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 52.4 TFLOPS. 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.
Clocks run from 900 MHz at base to 1.6 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 Data Center GPU Max Subsystem has 64 KB of L1 cache, backed by 408 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 16,384 shading units, 1,024 texture mapping 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
2,400 W
The Data Center GPU Max Subsystem is rated at 2,400 W, with a 2,800 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 dual-slot, measuring 267 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 Data Center GPU Max Subsystem 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 Data Center GPU Max Subsystem
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 Data Center GPU Max Subsystem
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
Search for the model you want
The table lists 624 models this Data Center GPU Max Subsystem 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
Longer conversations cost memory on top of the weights. With 128 GB to work in, that is frequently the difference between a model fitting and not.
-
03
Choose how far you will compress
By default the table picks the least-compressed copy that fits. Setting a floor removes models that only qualify through heavy compression.
-
04
Take the range as the answer
Speeds come with error bars for a reason. The best case here is 884 tok/s on Gemma 3 QAT 1B, and which inference software you use moves that by thirty to fifty per cent.
-
05
Check the headroom before you decide
Compare what each model needs with the 128 GB this card provides. Tight means it works today; comfortable means it still works when the conversation grows.
-
06
Check the same model from the other side
Following a model through to its own page lists all the hardware that can run it, so you can see where the Data Center GPU Max Subsystem sits against the alternatives.
Answers
Data Center GPU Max Subsystem — common questions
What are the TFLOPS of a Data Center GPU Max Subsystem?
The Data Center GPU Max Subsystem is rated at 52.4 TFLOPS at half precision and 52.4 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.
Does the Data Center GPU Max Subsystem support CUDA?
No. CUDA is NVIDIA-only, and the Data Center GPU Max Subsystem is a Intel card. It runs language models through ROCm, Vulkan or Metal depending on the software, which are less mature than the CUDA path — our estimates apply a penalty for that.
What bus interface does the Data Center GPU Max Subsystem 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 Data Center GPU Max Subsystem good for running local AI models?
Its memory is large enough for models most desktop hardware cannot touch and its bandwidth puts it among the fastest hardware available for generation. In total it runs 624 of the models we track. Whether that is enough depends entirely on which model you want — the table above answers that directly.
Can a Data Center GPU Max Subsystem run a model that does not fit in its memory?
Offloading past the card's 128 GB is possible and usually a false economy: the system-memory portion is slow enough to dominate the result.
Would two Data Center GPU Max Subsystem cards be twice as fast?
Pairing Data Center GPU Max Subsystem cards buys headroom rather than pace: 256 GB of combined memory, at roughly the same generation speed as one.
What AI models can a Data Center GPU Max Subsystem run?
624 of the 679 open-weight language models we track fit on a Data Center GPU Max Subsystem 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 Data Center GPU Max Subsystem can run?
The largest model in our catalogue that fits on a Data Center GPU Max Subsystem is Solar Open2 250B at 250.3B parameters, compressed to Q3_K_M. It generates roughly 52.9 tokens per second and needs about 106.6 GB of the card's memory.
How many tokens per second does a Data Center GPU Max Subsystem produce?
It depends on the model. On a Data Center GPU Max Subsystem the fastest model we track is Gemma 3 QAT 1B at about 884 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 Data Center GPU Max Subsystem run a 7B model?
Yes. For example a Data Center GPU Max Subsystem runs Multi-Token Prediction 7B at Q8_0, using about 7.9 GB of memory and generating around 132 tokens per second.
Can a Data Center GPU Max Subsystem run a 13B model?
Yes. For example a Data Center GPU Max Subsystem runs DeepSeekMoE-16B at Q8_0, using about 17.5 GB of memory and generating around 307 tokens per second.
Can a Data Center GPU Max Subsystem run a 30B model?
Yes. For example a Data Center GPU Max Subsystem runs ERNIE-4.5-VL-28B-A3B at Q8_0, using about 29.2 GB of memory and generating around 175 tokens per second.
Can a Data Center GPU Max Subsystem run a 70B model?
Yes. For example a Data Center GPU Max Subsystem runs Qwen3-Coder-Next at Q8_0, using about 80.7 GB of memory and generating around 61.4 tokens per second.
How much memory does a Data Center GPU Max Subsystem have?
A Data Center GPU Max Subsystem has 128 GB of HBM2e memory. Around a tenth of that is reserved by the inference runtime and the driver, leaving roughly 115.2 GB available for a model and its conversation.
What is the memory bandwidth of a Data Center GPU Max Subsystem?
The Data Center GPU Max Subsystem has 3,210 GB/s of memory bandwidth, across a 8,192-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 Data Center GPU Max Subsystem use?
It uses HBM2e clocked at 1.57 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 Data Center GPU Max Subsystem?
The Data Center GPU Max Subsystem is a Intel product, with the chip manufactured by Intel, on a 10 nm process.
When was the Data Center GPU Max Subsystem released?
The Data Center GPU Max Subsystem was released in January 2023.
How much power does a Data Center GPU Max Subsystem use?
The Data Center GPU Max Subsystem has a rated board power of 2,400 W, and a 2,800 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 Data Center GPU Max Subsystem have?
The Data Center GPU Max Subsystem has 64 KB of L1 cache, and 408 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.
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.