The compute market has two different questions: Is this resource intrinsically good? and Is this offer attractive right now? Treating them as one score creates bad recommendations. Compute Market Intelligence keeps them separate, then attaches evidence and freshness to the decision.
That distinction applies across VPS, GPU Cloud, AI API, and Credits. The fields differ by market, but the evaluation rule stays consistent: do not turn missing evidence into certainty.
Deal Score and Compute Score answer different questions
Deal Score describes the attractiveness of the current offer. It can change when price, terms, availability, or another verified deal condition changes.
Compute Score describes intrinsic resource quality. It should not rise merely because a provider runs a promotion, and it should not be invented when the evidence required to calculate it is unavailable.
A low-cost resource can therefore be a strong deal without being a strong machine. The reverse is also possible: a technically excellent resource may be a poor purchase at its current terms.
Decision rule: a current offer and the underlying resource are separate objects of judgment. Keep their scores separate before making a recommendation.
Fresh evidence beats stale certainty
A price observed yesterday and a price re-verified five minutes ago do not carry the same decision value. This is why CMI exposes checked time and treats freshness as part of the presentation layer rather than hiding it behind a score.
When a previously attractive offer becomes stale, the interface should reduce confidence in the recommendation instead of pretending the old state is still current. When newer verified evidence appears, it should take precedence over an older observation.
This article follows the same rule. The prose is based on the snapshot identified in the Data Basis panel. Resource cards shown later are resolved from current publishable data at build time, so they may be newer than the article itself.
Native price stays primary
Normalized monthly or unit costs are useful for comparison, but they are analytical aids. The provider-listed currency, billing period, and contract context remain the primary factual price.
This matters because a normalized number can hide setup charges, renewal terms, tax treatment, regional differences, minimum commitments, or eligibility conditions. CMI should never present a normalized number as though it were a universal checkout price.
The same caution applies to credits. An official credit value is not the same thing as acquisition cost, and a percentage discount should not be transformed into a made-up fixed cash value.
Evidence comes before interpretation
Price movement, availability, eligibility, expiry, specifications, and other decision-critical facts should remain traceable to source evidence. Analysis can explain why a change matters; it should not manufacture a fact that is absent from the underlying record.
This is also why a missing field is sometimes the correct output. If intrinsic quality cannot be calculated from the available data, showing “unknown” is more useful than filling the gap with a confident-looking number.
Commercial state is isolated from recommendation authority
Affiliate or sponsored status can determine which outbound link is eligible to be shown, but it must not determine which resource ranks first, which resource is published, or what Deal Score or Compute Score it receives.
That separation is deliberate. A commercial relationship can coexist with useful market intelligence only when the scoring and publication layers remain independent from the monetization layer.
Use analysis to understand the rule, then inspect current market state
An article is not a frozen price feed. Use analysis to understand the decision rule and the evidence behind it, then inspect the current market surface before acting. The VPS explorer is the right place to compare current virtual-server offers; GPU Cloud exposes current accelerator resources; AI API keeps model-access pricing in its native usage context; and Credits separates program benefit from acquisition cost and eligibility.
The resource references below are intentionally resolved from current publishable data. Their state can change without rewriting this historical analysis snapshot.