LLMBOTTLENECK.COM
OpenAI / gpt_oss

gpt-oss-120b

116.8 billion parameters, routing 4 of 128 experts per token.

Open in the calculator →

Architecturepublished data
Quick answer
gpt-oss-120b needs about 71.9 GB at Q4_K_M — the format most people download — with 8,192 tokens of context. It fits entirely on 5 of 18 common devices, starting with the AMD Ryzen AI Max+ 395 with Radeon 8060S (128 GB). What else fits in 128 GB. No hardware for it? Rent a GPU that runs it, priced live.

Sized at 8,192 tokens of context with the whole model in device memory, across 18 common devices. Speeds are estimates, not benchmarks.

Find the AMD Ryzen AI Max+ 395 with Radeon 8060S, the smallest card listed here that holds it: Amazon ↗ · eBay (new and used) ↗Store links may pay us a commission. They never decide which card is suggested — the memory arithmetic does.

Weight-file size by format

FormatSizeRangeBasis
FP16233.6 GB232.4 GB – 234.8 GBreconstructed size
Q8_0124.1 GB123.5 GB – 124.8 GBreconstructed size
Q6_K95.9 GB95.4 GB – 96.3 GBreconstructed size
Q5_K_M83.0 GB82.5 GB – 83.4 GBreconstructed size
Q5_080.4 GB79.9 GB – 80.8 GBreconstructed size
Q4_K_M70.8 GB70.3 GB – 71.2 GBreconstructed size
Q4_065.9 GB65.4 GB – 66.2 GBreconstructed size
Q3_K_M57.9 GB53.9 GB – 62.0 GBreconstructed size
Q2_K42.6 GB40.9 GB – 44.3 GBreconstructed size

A file size is not the memory a run needs: the KV cache and the runtime reserve come on top, and the calculator adds both.

What the labels mean
published data
Read from a published source — a file's byte count, a model's configuration or a manufacturer's specification — or exact arithmetic on such values. Not a measurement on a machine.
reconstructed size
Weight size reconstructed from the pinned architecture, because no published file exists.
size range
Only a lower and an upper bound are claimed for this weight size.

Hardware ladder

Q4_K_M at 8,192 tokens · 5 of 18 devices hold it entirely
DeviceMemoryVerdictDecodeCalculator
NVIDIA GeForce RTX 507071.9 GB of 12 GB31 layers on system RAM~21 tok/s with offloadOpen →
AMD Radeon™ RX 9070 XT71.9 GB of 16 GB29 layers on system RAM~22 tok/s with offloadOpen →
NVIDIA GeForce RTX 408071.9 GB of 16 GB29 layers on system RAM~22 tok/s with offloadOpen →
NVIDIA GeForce RTX 508071.9 GB of 16 GB29 layers on system RAM~22 tok/s with offloadOpen →
NVIDIA GeForce RTX 5070 Ti71.9 GB of 16 GB29 layers on system RAM~22 tok/s with offloadOpen →
NVIDIA GeForce RTX 5060 Ti 16GB71.9 GB of 16 GB29 layers on system RAM~21 tok/s with offloadOpen →
NVIDIA GeForce RTX 409071.9 GB of 24 GB25 layers on system RAM~25 tok/s with offloadOpen →
AMD Radeon™ RX 7900 XTX71.9 GB of 24 GB25 layers on system RAM~25 tok/s with offloadOpen →
NVIDIA GeForce RTX 309071.9 GB of 24 GB25 layers on system RAM~25 tok/s with offloadOpen →
NVIDIA GeForce RTX 509071.9 GB of 32 GB21 layers on system RAM~30 tok/s with offloadOpen →
NVIDIA RTX 6000 Ada Generation71.9 GB of 48 GB13 layers on system RAM~43 tok/s with offloadOpen →
Apple M4 Pro71.9 GB of 64 GBdoes not fitnot calibratedOpen →
Apple M5 Pro71.9 GB of 64 GBdoes not fitnot calibratedOpen →
AMD Ryzen AI Max+ 395 with Radeon 8060S71.9 GB of 128 GBfits~56 tok/sOpen →
Apple M3 Max71.9 GB of 128 GBfits~66 tok/sOpen →
Apple M4 Max71.9 GB of 128 GBfits~79 tok/sOpen →
NVIDIA DGX Spark71.9 GB of 128 GBfits~52 tok/sOpen →
Apple M2 Ultra71.9 GB of 192 GBfits~96 tok/sOpen →

Decode is a calibrated estimate from the published calibration; capacity is the manufacturer's published ceiling, not guaranteed free memory. Fitting in memory is not the same as loading: whether the runtime and version you have supports this architecture and format on that machine has not been tested here. Offloaded rows assume there is enough system RAM for the overflow — the calculator checks that against the RAM you declare. Each “Open” link keeps this model, format and context.

Run it in the cloud

Rent a GPU that runs gpt-oss-120b

Every rentable machine that holds the whole model, from Vast.ai and RunPod, with this site's speed estimate for each and the provider's own price, read live. Cheapest first; sort by cost per token or speed instead, or switch the format.

Prices · 18:30 UTC, 22 Sept
18 configurations in stock · 14 out of stock
MachineSpeedPer hourPer M tokensWhereRent
8× RTX 306096 GBcheapest~551 tok/sFaster than you read$0.59$0.30Vast.aiMarketplace · 99.7% reliableRent on Vast.ai
4× RTX 309096 GB~716 tok/sFaster than you read$0.60$0.23Vast.aiMarketplace · 99.4% reliableRent on Vast.ai
4× RTX 5090128 GBrecommendedbest value~1371 tok/sFaster than you read$1.07$0.22Vast.aiMarketplace · 99.6% reliableRent on Vast.ai
2× RTX 6000 Ada96 GB~367 tok/sFaster than you read$1.07$0.81Vast.aiMarketplace · 99.4% reliableRent on Vast.ai
1× RTX PRO 600096 GB~342 tok/sFaster than you read$1.14$0.92Vast.aiMarketplace · 98.8% reliableRunPod $1.69/h Rent on Vast.ai
1× A100 80GB PCIe80 GB~370 tok/sFaster than you read$1.19$0.89RunPodCommunity CloudRent on RunPod
2× RTX A600096 GB~293 tok/sFaster than you read$1.20$1.14Vast.aiMarketplace · 99.4% reliableRent on Vast.ai
4× L496 GB~229 tok/sFaster than you read$1.29$1.55Vast.aiMarketplace · 99.4% reliableRent on Vast.ai

No GPU at all

Use gpt-oss-120b by the token

Here the rented card is as cheap as the API or cheaper per token ($0.22 per million), and it keeps your data on a machine you control. The API is still the easier start: nothing to set up and nothing to switch off.

First time renting a GPU? How it works, in four steps
  1. Create an account and add credit. Both providers are prepaid: $10 is enough to try any card on this page for hours. A new RunPod account that arrives through a referral link gets a one-time $5 credit when it first adds $10.
  2. Pick the card and the count shown here, and a template: llama.cpp or vLLM for an OpenAI-compatible API, or one with a web chat if you only want to talk to the model. “How to launch it” below gives the exact image and command.
  3. Wait for the download. The weights are fetched on the machine; tens of gigabytes take a few minutes on a data-centre connection. You pay by the second from the moment it starts.
  4. Stop it when you are done. A running machine bills even when idle. A stopped one costs nothing per hour, though its disk is usually still billed until you delete it.
How to launch it

No Q4_K_M file of gpt-oss-120b is catalogued here, so there is no exact command to vouch for. Start the machine from the llama.cpp server image ghcr.io/ggml-org/llama.cpp:server-cuda and point -hf at a Q4_K_M build from the publisher or a quantizer you trust — search Hugging Face for one. Check its file size against the memory figure on this page before you rent.

Every machine holds the whole model at Q4_K_M and 8,192 tokens of context, no offload. Speeds are this site’s single-stream decode estimates; prices are what each provider’s own API quoted, on-demand, for the whole machine. Vast hosts below 98% measured reliability are left out.

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

Buying a card instead, or paying by the token? Compare a year of gpt-oss-120b three ways →

Under the hood

Architecture, read from the publisher’s file

The numbers every figure above is computed from, with the file they came from.

Architecture

✓ Architecture read from the published config.json

Retrieved 2026-09-01 at pinned commit b5c939de8f75; parameter count from the safetensors index at the same revision.

Architecture
gpt_oss
Layers
36
Hidden size
2,880
Attention heads
64
KV heads
8
Head dimension
64
Feed-forward width
2,880
Vocabulary
201,088
Context ceiling
131,072
Sliding window
128
RoPE theta
150,000
Experts
128
Experts / token
4

Where the memory goes

tokenembedding× 36 decoder blocksGrouped-query attention64 query · 8 KV headswindow 128Routed experts4 of 128 per tokenall residentoutputprojectiongrows with contextcapacity ≠ traffic

Grouped-query attention shares each key/value head across 8 query heads, so the KV cache is 13% of what multi-head attention would need at the same context.

Open the exact official config.json ↗

gpt-oss-120b, card by card

One page per device: whether gpt-oss-120b fits, at which formats, and how fast.

All 30Fewer

More from OpenAI

gpt-oss-20bgpt-oss-safeguard-20b

Use this answer in your own product

As JSON, from the API
curl -s https://llmbottleneck.com/v1/analyze \
  -H "Authorization: Bearer $LLMB_KEY" -H "content-type: application/json" \
  -d '{"model":"openai-gpt-oss-120b","quantization":"Q4_K_M","context":8192,"hardware":"nvidia-geforce-rtx-4090"}'

Same engine, same evidence, every field sourced. Free key in one step →

On your page, as a widget
<script src="https://llmbottleneck.com/widget.js"
  data-model="openai-gpt-oss-120b" data-quantization="Q4_K_M"
  data-hardware="nvidia-geforce-rtx-4090" data-context="8192"></script>

No key needed. Unbranded, with your own buy button, on Pro and Business →

Citing this page

LLM Bottleneck. “gpt-oss-120b VRAM and hardware requirements.” Architecture from openai/gpt-oss-120b at revision b5c939de8f75, retrieved 2026-09-01. https://llmbottleneck.com/models/openai-gpt-oss-120b

Every figure above is either the published value or a reconstruction whose measured error is on the accuracy page.

llmbottleneck
catalogue 2026-10-03models 327devices 135