OpenAI · 117B (5.1B active) parameters · Apache 2.0
What GPU do you need to run gpt-oss-120b?Memory, cards and speed.
Even at its smallest (MXFP4, as published), gpt-oss-120b needs about 64 GB of GPU memory: more than any single consumer card has.
GPU memory
gpt-oss-120b, by version.
Includes about 1.5 GB for the program and a short conversation (about 4,000 tokens). Long conversations need more. Speeds are ceilings: real speeds are usually half to two-thirds of them. How we work it out
| Version | Memory needed | Card size | Cards that fit | From | Top speed |
|---|---|---|---|---|---|
| MXFP4, as published | about 64 GB | None | More than any consumer card holds | — | Faster than its size suggests |
OpenAI's larger open-weight model. OpenAI says it runs on a single 80 GB GPU such as an H100; no consumer card holds it.
Run it
Too big for one card here, for now.
gpt-oss-120b needs about 64 GB even at its smallest, and the biggest card on KernelMachine has 32 GB. Ollama would run most of it from the computer's memory instead, many times slower, so a smaller model is the better pick here. On a machine with enough GPU memory, one command downloads and runs it. Models that fit
$ ollama run gpt-oss:120b
# Model files: huggingface.co/openai/gpt-oss-120b
Common questions
Questions, answered plainly.
How much VRAM does gpt-oss-120b need?
MXFP4, as published: about 64 GB. That includes about 1.5 GB for the program and a short conversation; longer conversations need more.
Can an RTX 4090 run gpt-oss-120b?
Not on its own. Even at its smallest, gpt-oss-120b needs about 64 GB, more than the RTX 4090's 24 GB.
How fast does gpt-oss-120b run?
Faster than its size suggests. It's a mixture-of-experts model, so it reads only about 5.1B of its 117B parameters for each word.
Can I run gpt-oss-120b on KernelMachine?
Not yet. KernelMachine rents one card at a time, and the biggest has 32 GB; gpt-oss-120b needs about 64 GB even at its smallest.
Other models
Compare with other chat models.
Preparing an invite-only beta
Get an email when access opens.
No launch date is set. We'll email you once when KernelMachine opens. Nothing else.