Nova AI

Nova Premier (80B)

Latest Nova high-end model for complex reasoning tasks

Model Summary

Family

Nova

Version

2.0

Parameters

80B (est.)

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

VRAM Requirements by Quantization

Memory needed to serve Nova Premier (80B) for inference, including a 20% overhead for activations and KV cache.

PrecisionVRAM neededSmallest single GPU that fits
INT4 (4-bit)44.70 GBNVIDIA L40S (48 GB)
INT8 (8-bit)89.41 GBNVIDIA RTX PRO 6000 Blackwell (96 GB)
FP16 (16-bit)178.81 GBNVIDIA B200 SXM (192 GB)
FP32 (32-bit)357.63 GBMulti-GPU required

Recommended GPU Configurations

Cheapest on-demand configurations to serve Nova Premier (80B) at 8-bit (89 GB VRAM).

1x AMD Instinct MI250X

128 GB total VRAM · CDNA 2

~$2.50/h

6x NVIDIA T4

96 GB total VRAM · Turing

~$3.00/h

6x NVIDIA P100 SXM2

96 GB total VRAM · Pascal

~$3.60/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 Nova Premier (80B)?

With an estimated 80B parameters, Nova Premier (80B) needs roughly 89 GB of VRAM in 8-bit (INT8), 45 GB in 4-bit, and 179 GB in FP16, including a 20% overhead for activations and KV cache.

Which GPUs can run Nova Premier (80B)?

At 8-bit quantization, the most cost-effective option is 1x AMD Instinct MI250X (128 GB combined VRAM, around $2.50/hour on-demand). Higher-end cards like the NVIDIA B200 or AMD MI355X reduce the GPU count needed.

Can Nova Premier (80B) run on a single GPU?

Yes. In 8-bit, a single NVIDIA RTX PRO 6000 Blackwell (96 GB) fits the model.

How much does it cost to serve Nova Premier (80B) in the cloud?

Renting 1x Instinct MI250X costs on the order of $2.50/hour, i.e. about $1,825/month running 24/7. Actual prices vary by provider and commitment; spot and reserved capacity can be significantly cheaper.

Other Nova Versions