GPU comparison · live prices
RTX PRO 6000 vs 4090: price per hour and cost per token
Blackwell workstation and server card: more memory than an H100, usually for less per hour, but no NVLink, so best when a model fits on one. The cheapest way to serve 7B–14B models, mostly on marketplaces and spot.
Prices observed 13m ago · refreshed every 15 minutes
| NVIDIA RTX PRO 6000 | NVIDIA RTX 4090 | |
|---|---|---|
| Memory | 96 GB GDDR7 | 24 GB GDDR6X |
| Memory bandwidth | 1.8 TB/s | 1.0 TB/s |
| Cheapest on-demandper GPU-hour | $1.00Vast.ai | $0.31Vast.ai |
| Cheapest spotper GPU-hour | $1.00Nebius · availability not published | —None in stock |
| Providers in stock | 3 | 3 |
Per GPU-hour on demand, the 4090 costs 0.31× the RTX PRO 6000. Per token, that depends on the model:
Cost per 1M output tokens
Cheapest on-demand offer, GPUs per replica to fit the model, 60% average utilization.
RTX PRO 6000 is cheaper per token for 1 of these models and 4090 for 4: it depends on model size.
| Model | RTX PRO 6000 | 4090 |
|---|---|---|
| Llama 3.1 8BBF16 | $0.17 1× GPUs · ~2,688 tok/s | $0.10 1× GPUs · ~1,512 tok/s |
| Qwen 2.5 32BFP8 | $0.35 1× GPUs · ~1,323 tok/s | $0.32 2× GPUs · ~893 tok/s |
| Llama 3.3 70BFP8 | $0.76 1× GPUs · ~609 tok/s | $0.70 4× GPUs · ~822 tok/s |
| Mixtral 8x22BFP8 | $0.70 2× GPUs · ~1,323 tok/s | $0.39 8× GPUs · ~2,977 tok/s |
| Llama 3.1 405BFP8 | $7.28 8× GPUs · ~510 tok/s | — Doesn’t fit on 8 |
| DeepSeek V3 (671B MoE)FP8 | — Doesn’t fit on 8 | — Doesn’t fit on 8 |
Estimates: tokens/s from memory bandwidth with batched serving, not a benchmark. Check each provider before you buy.