Calculate the TPS of the Radeon Instinct MI350X 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
721 models in our catalogue altogether
Largest model it holds
Nemotron 3 Ultra
550B · Q3_K_M · 73.8 tok/s
Fastest model
Gemma 3 QAT 1B
2,706 tok/s · 1B
Which AI models can run on a Radeon Instinct MI350X?
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.
689 models match
Calculating| Quantisation | Fit | ||||||
|---|---|---|---|---|---|---|---|
|
2,706
tok/s
1,623–4,329 · low confidence |
Gemma 3 1B | 1B | Mar 2025 | 1.8 GB | 33k tokens | Q8_0 | Comfortable |
|
2,706
tok/s
1,623–4,329 · low confidence |
Gemma 3 QAT 1B | 1B | Apr 2025 | 1.8 GB | 33k tokens | Q8_0 | Comfortable |
|
2,706
tok/s
1,623–4,329 · low confidence |
HGRN 1B (WT 103) ≈ | 1B | Nov 2023 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,706
tok/s
1,623–4,329 · low confidence |
LLama 3..2 Typhoon 2 1B ≈ | 1B | Dec 2024 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,706
tok/s
1,623–4,329 · low confidence |
OLMo-1B ≈ | 1B | Feb 2024 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,706
tok/s
1,623–4,329 · low confidence |
Pythia-1b ≈ | 1B | Apr 2023 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,505
tok/s
1,503–4,008 · low confidence |
OpenELM-1.1B ≈ | 1.1B | May 2024 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,460
tok/s
1,476–3,935 · low confidence |
DeciCoder-1B ≈ | 1.1B | Aug 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,460
tok/s
1,476–3,935 · low confidence |
SantaCoder ≈ | 1.1B | Jan 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,460
tok/s
1,476–3,935 · low confidence |
TinyLlama-1.1B (1T token checkpoint) ≈ | 1.1B | Oct 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,460
tok/s
1,476–3,935 · low confidence |
TinyLlama-1.1B (3T token checkpoint) ≈ | 1.1B | Oct 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,255
tok/s
1,353–3,607 · low confidence |
EXAONE 4.0 (1.2B) ≈ | 1.2B | Jul 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,255
tok/s
1,353–3,607 · low confidence |
LFM2-1.2B ≈ | 1.2B | Jul 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,255
tok/s
1,353–3,607 · low confidence |
MinerU2.5 ≈ | 1.2B | Sep 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,255
tok/s
1,353–3,607 · low confidence |
Pleias 1.0 1.2B ≈ | 1.2B | Dec 2024 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,255
tok/s
1,353–3,607 · low confidence |
Pleias-RAG-1B ≈ | 1.2B | Apr 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,200
tok/s
1,320–3,519 · low confidence |
Llama 3.2 1B | 1.2B | Sep 2024 | 2.2 GB | 131k tokens | Q8_0 | Comfortable |
|
2,169
tok/s
1,301–3,470 · low confidence |
MiniCPM-1.2B ≈ | 1.2B | Jun 2024 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,081
tok/s
1,249–3,330 · low confidence |
DeepSeek Coder 1.3B ≈ | 1.3B | Jan 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,081
tok/s
1,249–3,330 · low confidence |
DeepSeek-VL-1.3B ≈ | 1.3B | Mar 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,081
tok/s
1,249–3,330 · low confidence |
DigiRL ≈ | 1.3B | Jun 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,081
tok/s
1,249–3,330 · low confidence |
GLA Transformer 1.3B ≈ | 1.3B | Aug 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,081
tok/s
1,249–3,330 · low confidence |
Janus 1.3B ≈ | 1.3B | Oct 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,081
tok/s
1,249–3,330 · low confidence |
Kosmos-2.5 ≈ | 1.3B | Aug 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
2,081
tok/s
1,249–3,330 · low confidence |
Otter ≈ | 1.3B | May 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
Radeon Instinct MI350X 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
- 288 GB
- Memory bandwidth
- 8,190 GB/s
- Memory type
- HBM3e
- Memory bus width
- 8,192 bit
- Memory clock
- 2 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
- Galaxy
- Architecture
- CDNA 4.0
- Generation
- Radeon Instinct(MIx)
- Foundry
- TSMC
- Process size
- 3 nm
- Released
- 1 January 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 GHz
- Boost clock
- 2.2 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
- L1 cache
- 16 KB
- L2 cache
- 16 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)
- 576.7 TFLOPS
- Single precision (FP32)
- 72.1 TFLOPS
- Double precision (FP64)
- 72.1 TFLOPS
- Texture rate
- 2,253 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)
- 1,000 W
- Suggested power supply
- 1,400 W
- Power connectors
- None
- Bus interface
- PCIe 5.0 x16
- Slot width
- OAM Module
- Dimensions
- 102 mm × 165 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.
- OpenCL
- 3.0
Listings
Where to buy a Radeon Instinct MI350X
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
Why memory is the number that matters here
Memory
288 GB
Bandwidth
8,190 GB/s
Largest model
Nemotron 3 Ultra
Radeon Instinct MI350X holds 288 GB of HBM3e. That puts it in the class of hardware that holds the largest open-weight models without splitting them across machines. An inference runtime can reach roughly 259.2 GB.
Memory bandwidth reaches 8,190 GB/s across a bus of 8,192 bits. That is at the top of what exists. Since each token means reading the whole model out of memory once, it 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.
Bandwidth is clock times bus width, and this card clocks its memory at 2 GHz. Both halves matter, and neither is visible in a gaming benchmark.
The biggest thing it holds is Nemotron 3 Ultra, 550B, compressed to Q3_K_M and generating around 73.8 tokens per second.
The chip and how it was built
Radeon Instinct MI350X is built on the graphics processor Galaxy, using the architecture CDNA 4.0 from AMD, as part of the generation Radeon Instinct(MIx).
The chip is manufactured by TSMC, on a process of 3 nm. 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 2025, roughly 1.7006419936991 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
576.7 TFLOPS
FP64
72.1 TFLOPS
On paper Radeon Instinct MI350X reaches 576.7 TFLOPS at half precision, and 72.1 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 reaches 72.1 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 a base of 1 GHz to a boost of 2.2 GHz. 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
Radeon Instinct MI350X has an L1 cache of 16 KB, backed by an L2 cache of 16 MB. 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
1,000 W
Radeon Instinct MI350X is rated at 1,000 W, and the suggested system power supply is 1,400 W. 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 oam module, measuring 102 mm long. 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 that run on a Radeon Instinct MI350X
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 Radeon Instinct MI350X
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 Radeon Instinct MI350X
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
Find the model in the table
The table lists 689 models this card runs. Search narrows the list by name or by size.
-
02
Decide how long your conversations run
Longer conversations cost memory on top of the weights. Against 288 GB that is frequently the difference between a model fitting and not.
-
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
Look at the range, not just the number
Each speed is an estimate for a single conversation, with a range beneath it. The top end here is 2,706 tok/s on Gemma 3 QAT 1B. The same card and model vary by thirty to fifty per cent between inference runtimes.
-
05
Read the fit verdict last
The fit column separates models that just fit from those with room to spare — worth checking before settling on one, against an available 288 GB.
-
06
Open the model to compare cards
Each model page repeats this calculation for the whole catalogue. Worth a look before deciding: it shows what else runs the same model, alongside Radeon Instinct MI350X.
Answers
Radeon Instinct MI350X — common questions
Radeon Instinct MI350X— can it run 7B models?
Yes. For example it runs Gemma 4 E4B at Q8_0, using about 9.8 GB of memory and generating around 601 tokens per second.
Radeon Instinct MI350X— can it run 13B models?
Yes. For example it runs DeepSeekMoE-16B at Q8_0, using about 17.5 GB of memory and generating around 939 tokens per second.
Radeon Instinct MI350X— can it run 30B models?
Yes. For example it runs ERNIE-4.5-VL-28B-A3B at Q8_0, using about 29.2 GB of memory and generating around 537 tokens per second.
Radeon Instinct MI350X— can it run 70B models?
Yes. For example it runs Qwen3-Coder-Next at Q8_0, using about 80.7 GB of memory and generating around 188 tokens per second.
Radeon Instinct MI350X— how much memory does it have?
This card has 288 GB of HBM3e. Around a tenth is reserved by the inference runtime and the driver, leaving roughly 259.2 GB available for a model and its conversation.
Radeon Instinct MI350X— what is its memory bandwidth?
Memory bandwidth reaches 8,190 GB/s across a bus of 8,192 bits. 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.
Radeon Instinct MI350X— what type of memory does it use?
It uses HBM3e clocked at 2 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.
Radeon Instinct MI350X— who makes it?
This is a product of AMD, with the chip manufactured by TSMC, on a process of 3 nm.
Radeon Instinct MI350X— when was it released?
It was released in January 2025.
Radeon Instinct MI350X— how much power does it use?
Rated board power is 1,000 W, and the suggested system power supply is 1,400 W. Generating text draws hard in bursts and idles between requests, so average consumption over a working session is normally well below the rated figure.
Radeon Instinct MI350X— how much cache does it have?
The L1 cache is 16 KB, and the L2 cache is 16 MB. 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.
Radeon Instinct MI350X— what are its TFLOPS?
It is rated at 576.7 TFLOPS at half precision and 72.1 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.
Radeon Instinct MI350X— does it support CUDA?
No. CUDA is NVIDIA-only, and this is a card from AMD. 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.
Radeon Instinct MI350X— what bus interface does it 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.
Radeon Instinct MI350X— is it 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 689 of the models we track. Whether that is enough depends entirely on which model you want — the table above answers that directly.
Radeon Instinct MI350X— can it run a model that does not fit in its memory?
It can be split, with the overflow held in system memory beyond the card's 288 GB is possible and usually a false economy: the system-memory portion is slow enough to dominate the result.
Would two Radeon Instinct MI350X cards be twice as fast?
Capacity adds, throughput does not. Two of them give you 576 GB which lets you hold models neither could hold alone, but generation does not split that way. These figures describe one card.
Radeon Instinct MI350X— which AI models can it run?
689 of the 721 open-weight language models we track fit on this card and can be run locally. The table on this page lists every one, with the memory it needs, the quantisation it runs at and an estimated generation speed.
Radeon Instinct MI350X— what is the largest AI model it can run?
The largest model in our catalogue that fits is Nemotron 3 Ultra at 550B parameters, compressed to Q3_K_M. It generates roughly 73.8 tokens per second and needs about 239.6 GB of the card's memory.
Radeon Instinct MI350X— how many tokens per second does it produce?
It depends on the model. The fastest model we track here is Gemma 3 QAT 1B at about 2,706 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.
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.