Together launches preemptible compute for GPU clusters at 50% of on-demand price
togethercompute · x · 2026-09-11
Together AI announced public preview of preemptible compute for Together GPU Clusters:
- Pricing: same NVIDIA GPU infrastructure as on-demand, flat at 50% of on-demand rate with sub-hourly billing — no spot-market volatility.
- Mechanics: preemptible nodes draw from unused capacity and can be reclaimed. Reclamation triggers up to a five-minute drain window: the node is cordoned, pods receive SIGTERM, and workloads get up to five minutes to checkpoint and exit. The cluster automatically refills toward its preemptible target.
- Use cases: evals, fine-tuning, batch inference, and short interruption-tolerant experiments; standard nodes remain synchronously allocated and never preempted.
- Available today on Kubernetes clusters in all regions, on new or existing clusters.
A practical option for teams trading interruption tolerance for a 50% cost cut.
More from Infra
- SemiAnalysis: OpenAI reportedly negotiating ~1GW TPU commitment while seeking sites — dylan522p · 2026-10-28
- I tested 11 video upscaling methods for AI video: local, Topaz and cloud compared — lopas99 · 2026-10-28
- Thermo-backed CTO says AI is in its dial-up phase, touts disaggregated compute for an ultrafast 'broadband moment' — beffjezos · 2026-10-28
- DeepSeek-V4.1-Flash Q2 runs at 41 tok/s on one RTX 3090, 90% token match vs API — musicurgy · 2026-10-28
- USearch v3 RC ships after 1,000 days, its libraries now embedded in 550K-star projects — viglovikov · 2026-10-28
- SK Hynix says memory demand will outpace supply through 2030 — Chemical_Hawk_6307 · 2026-10-28