GPU Servers for LLM Training
Compare GPU rentals suited to large language model training by GPU memory, GPU count, system RAM and hourly price.
At the latest catalog refresh, the lowest listed matching offer is $0.253 / hour from Vast.ai in , US. Availability and final checkout pricing can change after this timestamp.
Compare the full configuration, not only the headline price
ComputeRadar uses transparent minimum hardware filters for this workload page rather than claiming that every listed server is automatically optimal for every implementation.
Treat these results as a shortlist. Software stack, model size, concurrency, storage, network and provider-specific features can change the best choice.
What to check before renting
Prioritize GPU interconnect and multi-GPU topology for distributed training.
Check whether storage and CPU are included in the displayed GPU rate.
Large models may benefit from 80 GB+ VRAM per GPU.
Common questions
How are LLM training offers selected?
The page applies the hardware constraints shown in its description to the same normalized live catalog used by the main search.
Is this a benchmark ranking?
No. ComputeRadar is comparing available infrastructure and pricing, not claiming application-level performance without benchmark evidence.