ComparisonHost

Llama 3.3 70B

by Metaopen weights

API Pricing

$1.04
per 1M input tokens
$1.04
per 1M output tokens

Open weights; price shown is Together AI serverless hosting. Context per Meta model card.

Verified 2026-07-14 · official pricing page

Specifications

Context window128K tokens
Modalitiestext
Parameters70B

Run It Yourself?

Llama 3.3 70B ships open weights (70B parameters). Rough VRAM need — computed as parameters × bytes-per-weight × 1.2 overhead: ~168GB at FP16 or ~42GB with 4-bit quantization.

Cheapest live GPU servers clearing the 4-bit bar (42GB):

Multi-GPU Dedicated Server - 2xRTX A4000DatabaseMart · 2 x Nvidia RTX A4000 · 32GB ×2
$419.00/moGet server →
Multi-GPU Dedicated Server - 3xRTX A5000DatabaseMart · 3 x Quadro RTX A5000 · 72GB ×3
$454.35/moGet server →
Multi-GPU Dedicated Server - 3xRTX 3060 TiDatabaseMart · 3 x GeForce RTX 3060 Ti · 24GB ×3
$499.00/moGet server →