xAI

Grok 4.5

xAI's coding-focused model with enhanced agentic quality

Model Summary

Family

Grok

Version

4.5

Parameters

350B (est.)

Parameter counts for closed models are estimates; vendors rarely publish exact sizes.

VRAM Requirements by Quantization

Memory needed to serve Grok 4.5 for inference, including a 20% overhead for activations and KV cache.

PrecisionVRAM neededSmallest single GPU that fits
INT4 (4-bit)195.58 GBAMD Instinct MI325X (256 GB)
INT8 (8-bit)391.16 GBMulti-GPU required
FP16 (16-bit)782.31 GBMulti-GPU required
FP32 (32-bit)1564.62 GBMulti-GPU required

Recommended GPU Configurations

Cheapest on-demand configurations to serve Grok 4.5 at 8-bit (391 GB VRAM).

2x AMD Instinct MI325X

512 GB total VRAM · CDNA 3

~$16.00/h

3x AMD Instinct MI300X

576 GB total VRAM · CDNA 3

~$18.00/h

5x NVIDIA A100 80GB

400 GB total VRAM · Ampere

~$22.50/h

Quick GPU Planning

Use the calculator pre-filled with this exact version to estimate memory, speed, and compute requirements in a few clicks.

Access Pre-filled Calculator

Frequently Asked Questions

How much VRAM do you need to run Grok 4.5?

With an estimated 350B parameters, Grok 4.5 needs roughly 391 GB of VRAM in 8-bit (INT8), 196 GB in 4-bit, and 782 GB in FP16, including a 20% overhead for activations and KV cache.

Which GPUs can run Grok 4.5?

At 8-bit quantization, the most cost-effective option is 2x AMD Instinct MI325X (512 GB combined VRAM, around $16.00/hour on-demand). Higher-end cards like the NVIDIA B200 or AMD MI355X reduce the GPU count needed.

Can Grok 4.5 run on a single GPU?

Only with aggressive quantization: in 4-bit, a single AMD Instinct MI325X (256 GB) can fit it. At 8-bit or higher, you need a multi-GPU setup.

How much does it cost to serve Grok 4.5 in the cloud?

Renting 2x Instinct MI325X costs on the order of $16.00/hour, i.e. about $11,680/month running 24/7. Actual prices vary by provider and commitment; spot and reserved capacity can be significantly cheaper.

Other Grok Versions