Question 1

How much VRAM does gpt-oss-puzzle-88B need?

Accepted Answer

gpt-oss-puzzle-88B (90.8B) requires approximately 203.03835797309875 GB of VRAM at full precision (FP16), 101.5 GB at 8-bit quantization (Q8), and 50.8 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-puzzle-88B?

Accepted Answer

We have 2 compatible GPUs indexed for gpt-oss-puzzle-88B, including MI325X, 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-puzzle-88B?

Accepted Answer

We don't currently have cloud pricing data for GPUs compatible with gpt-oss-puzzle-88B. Cloud pricing data is updated regularly — check back soon.

Question 4

Can gpt-oss-puzzle-88B run on consumer hardware?

Accepted Answer

gpt-oss-puzzle-88B requires approximately 203.03835797309875 GB of VRAM at FP16, and even with 4-bit quantization (Q4) it needs about 50.8 GB. This exceeds the 24 GB available on the highest-end consumer GPU (RTX 4090), so gpt-oss-puzzle-88B requires datacenter or cloud GPUs.

Question 5

What quantization options are available for gpt-oss-puzzle-88B?

Accepted Answer

gpt-oss-puzzle-88B supports the following quantization levels: FP16 (full precision): ~203.03835797309875 GB, Q8 (8-bit): ~101.5 GB, Q6 (6-bit): ~76.1 GB, Q4 (4-bit): ~50.8 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.

gpt-oss-puzzle-88B

VRAM Requirements

Compatible GPUs

gpt-oss-puzzle-88B FAQ

VRAM requirements for gpt-oss-puzzle-88B at different quantization levels
Quantization	VRAM Required
FP16	203 GB
Q8	101.5 GB
Q6	76.1 GB
Q4	50.8 GB

Top 10 compatible GPUs for gpt-oss-puzzle-88B, sorted by cheapest price
GPU	VRAM	Best Quant	From	Rent
MI325X	256 GB	FP16	—
MI350X	288 GB	FP16	—