NVIDIA GeForce RTX 4090: specs, pricing and what it can run
Still the price/performance reference for local LLM work, though 24GB caps it at ~30B parameters in 4-bit.
Specifications
| Spec | NVIDIA GeForce RTX 4090 |
|---|---|
| Architecture | Ada Lovelace |
| Released | 2022 |
| Memory | 24 GB GDDR6X |
| Memory bandwidth | 1,008 GB/s |
| Dense FP16 | 330 TFLOPS |
| FP8 tensor cores | Yes |
| Board power | 450 W |
| Interconnect | PCIe Gen4 (no NVLink) |
| Segment | consumer |
What fits in 24 GB
Weights only, at 85% of nominal capacity to leave room for the KV cache and activations. Long-context serving needs materially more headroom than this table implies.
| Quantisation | Largest model (weights only) |
|---|---|
| FP16 | ~10B parameters |
| FP8 / INT8 | ~21B parameters |
| 4-bit | ~43B parameters |
Rental cost
Indicative on-demand pricing runs roughly $0.30–$0.90 per GPU-hour depending on provider, region, and commitment. Spot and reserved capacity sit well below that band. These are ranges rather than quotes: street prices move week to week, so check retail + live cloud price feed before budgeting.