Question 1

How much VRAM does gpt-oss-120b need?

Accepted Answer

gpt-oss-120b (120.4B) requires approximately 261.04386591911316 GB of VRAM at full precision (FP16), 130.5 GB at 8-bit quantization (Q8), and 65.3 GB at 4-bit quantization (Q4). The actual VRAM usage may be slightly higher due to KV cache and runtime overhead.

Question 2

What GPUs can run gpt-oss-120b?

Accepted Answer

We have 1 compatible GPUs indexed for gpt-oss-120b, including MI350X. Compatibility is determined by whether the GPU's VRAM can fit the model's memory footprint, potentially using quantization to reduce memory usage.

Question 3

What is the cheapest GPU to run gpt-oss-120b?

Accepted Answer

We don't currently have cloud pricing data for GPUs compatible with gpt-oss-120b. Cloud pricing data is updated regularly — check back soon.

Question 4

Can gpt-oss-120b run on consumer hardware?

Accepted Answer

gpt-oss-120b requires approximately 261.04386591911316 GB of VRAM at FP16, and even with 4-bit quantization (Q4) it needs about 65.3 GB. This exceeds the 24 GB available on the highest-end consumer GPU (RTX 4090), so gpt-oss-120b requires datacenter or cloud GPUs.

Question 5

What quantization options are available for gpt-oss-120b?

Accepted Answer

gpt-oss-120b supports the following quantization levels: FP16 (full precision): ~261.04386591911316 GB, Q8 (8-bit): ~130.5 GB, Q6 (6-bit): ~97.9 GB, Q4 (4-bit): ~65.3 GB. Lower quantization reduces VRAM usage at the cost of some quality loss. Q8 offers a good balance between quality and memory savings, while Q4 provides maximum compression. As a Mixture of Experts (MoE) model with 128 experts (4 active), gpt-oss-120b has a larger total weight size but only activates a subset of parameters per token, which can affect effective memory usage during inference.

gpt-oss-120b

VRAM Requirements

Compatible GPUs

gpt-oss-120b FAQ

VRAM requirements for gpt-oss-120b at different quantization levels
Quantization	VRAM Required
FP16	261 GB
Q8	130.5 GB
Q6	97.9 GB
Q4	65.3 GB