GPU watch planner // workload before SKU

Build a GPU watchlist around the model you run.

Choose a model, quantization, and budget—or watch one GPU directly. GPU Hunter will preview the indexed hardware that fits. Reference prices are dated estimates, not live offers.

Watch planner beta · no live offer feed
Loading watch planner…

Track a workload, not only a SKU

Workload mode starts from model memory and an explicit operating buffer. That keeps a tempting reference price from outranking the basic requirement that the hardware can load what you intend to run.

Specific-GPU mode is available when you already know the device you want. Both modes generate shareable criteria and can save the watch locally without collecting an email address or claiming monitoring has started.

What verified alerts must prove

  • Exact GPU and board identity with a visible confidence rule.
  • Condition, seller, shipping, and total delivered price.
  • Provider timestamps and a strict stale-offer cutoff.
  • Direct provider URLs and required affiliate disclosures.
  • Source-specific deletion and historical-retention behavior.

Source requirements before launch

GPU Hunter will not commit to a provider order or launch date until official access and commercial terms are confirmed. Every enabled source must support exact hardware identity, condition and delivered price, a visible observation time, enforceable stale-data deletion, required attribution, and explicit permission for the planned alert behavior.

Frequently asked questions

Are GPU Hunter price alerts live?

No. This beta builds, saves, and shares a watch plan from GPU Hunter's dated reference index. Automated price, stock, and email alerts will remain disabled until verified provider feeds and delivery infrastructure are connected.

Does saving a watch start retailer monitoring?

No. Saved watches stay in this browser and do not create a server-side subscription. The page will never label a watch active until GPU Hunter can verify offer identity, condition, price freshness, and delivery status.

Why build a watchlist around an AI model?

A cheaper GPU is not useful when it cannot load the model and precision you need. Workload mode adds planning headroom, filters the reference index by memory capacity, and then compares the remaining hardware with your target budget.

Are used, open-box, and new GPU prices combined?

No live offers are present in this version. Condition is saved as future alert intent but does not filter the dated reference dataset, because those reference values are not a condition-specific marketplace feed.

How will GPU Hunter choose offer sources?

A source can launch only after GPU Hunter verifies official access, exact GPU matching, freshness limits, attribution, retention, affiliate obligations, and alert permissions. No provider order or launch date is committed yet.