NVIDIA
GeForce RTX 3090 24GB
New · GPU only · host PC extra
Price unknown
No fresh confirmed in-stock offer right now.
USD. Retailer checkout determines shipping, tax, exact configuration and final availability. We may earn a commission from retailer links.
NVIDIA GeForce RTX 3090 Founders Edition shown as an example card. Partner-card designs may differ; the photo does not establish an offer MPN.
Photo: Adam Kapetanakis (User:Kaptainvet) · Source · CC BY-SA 4.0
None; original file copied without edits.
Hardware at a glance
- Chip
- RTX 3090
- Memory
- 24 GB
- Memory type
- GDDR6X
- Memory bandwidth
- 936 GB/s
- Storage
- Not specified / GPU only
- Catalog release status
- available
Nominal memory and bandwidth are specifications, not usable model memory or measured inference speed. Release status does not establish retailer stock.
Manufacturer specificationsAll retailer offers · New
A low price qualifies for the primary action only with fresh evidence and confirmed online stock. Unknown stock stays unconfirmed.
No display-permitted retailer offers are available. Stock is unknown. Set an alert below or check the manufacturer’s specifications.
Check Micro Center directly
Micro Center is link-only here. We do not display its price, store stock or product photos. Confirm the exact configuration and selected store on its website.
No verified Micro Center listing is linked to this configuration yet.
Stores & data checked
Coverage is limited to the permitted source records below for this configuration. Check times describe source observations; a daily feed is not a live store inventory check.
No permitted retailer price or stock records were returned. Availability at unlisted stores is unknown.
Micro Center links are checked by you on the retailer’s site. A store without a displayed record has no confirmed price or stock here.
Price unknown
New · No confirmed offer
Software & configuration checks
Guidance checked 2026-10-22 · verify current driver and runtime versions
Windows · Linux; check the driver and runtime matrix
CUDA / llama.cpp / Ollama
Check GPU architecture, driver minimum and runtime version; available kernels and quantizations vary.
Reference board power: 350 W. Partner cards and PSU requirements can differ.
- Check the exact partner card: board power, PSU rating and required power connectors.
- Check card length, height, occupied slots, cable clearance and cooling airflow.
- Check motherboard PCIe lanes and slots; multiple GPUs need an explicitly supported runtime and model split.
- Memory capacity does not certify model support or inference speed. Check the model, quantization, context and runtime version.
- Multiple GPUs or machines are not a transparent pooled-memory device; splitting support and link bandwidth matter.
- Use free VRAM after display and runtime allocations; host RAM does not replace dedicated VRAM at the same speed.
- The card price excludes the host PC, system RAM, storage, PSU and cooling upgrades.
Estimate inference memory
Use the total model parameters, including all MoE experts. Enter measured free accelerator memory after OS and application use, in one supported memory pool. Catalog capacity is not usable memory.
Observed price history · New
Not enough retained observations to show a trend yet. Tracking starts when a permitted source records prices; no long-term low or buying verdict is available.
Set a stock or price alert
Get notified about a fresh confirmed in-stock offer for GeForce RTX 3090 24GB. Leave the target blank for a stock alert.
Alert condition: New
Alerts cover eligible stores only; some website offers are not available by email. Double opt-in required. An alert becomes active only after you confirm your email and this condition. Changing this page does not change an existing alert. Each alert email includes an unsubscribe link.
Need help with model, quantization, context or runtime choices? Ask about your build