DocsICP FilingConsole
Documentation
Home Page Alibaba Cloud Model Studio User Guide (Agent) Memory Reference Billing

Billing

Updated at:
Copy as MD
Product
My Favorites

Commercial pricing, free quota, and billing examples

WarningMemory will begin commercial billing on August 20, 2026, 10:00 AM (Beijing Time).

Free quota

Free quota granted after commercialization:

ItemTiersPer-tier quotaTotalValidity
Memory write (Add)6 tiers250 calls/tier1,500 calls3 months, unused quota expires
Memory search (Search)2 tiers2,500 calls/tier5,000 calls3 months, unused quota expires
Memory storage——10,000 entriesNo expiry

NoteAdd's 6 tiers: Factual Pro, Factual Lite, Factual Multimodal Pro, Profile Pro, Profile Lite, Skill Pro. Search's 2 tiers: Pro, Lite.

Free quota offsets the corresponding billing item. Once exhausted, standard pricing applies.

Pricing

ItemPlan VersionPriceBillingNotes
Write - factual memoryPro¥0.03/callPer callIncludes inference and vectorization
Write - factual memoryLite¥0.018/callPer callSame
Write - factual memory (multimodal)Pro¥0.1/callPer callPro only, includes inference and vectorization
Write - profilesPro¥0.03/callPer callIncludes inference and vectorization
Write - profilesLite¥0.025/callPer callSame
Write - skill memoryPro¥0.3/callPer callPro only, includes inference and vectorization
SearchPro¥0.001/callPer callIncludes query vectorization and Rerank
SearchLite¥0.00002/callPer callIncludes vectorization, no Rerank
Storage—¥0.002/10K entries/hourBy duration~¥1.44/10K entries/month, billed hourly

Pro vs Lite

The core difference is whether Rerank is enabled during retrieval:

Plan VersionRerankQualityUse case
ProEnabledHigher, reranking model refines resultsAccuracy-critical scenarios
LiteDisabledStandard, skips rerankingHigh-frequency, cost-sensitive scenarios

Version control rules:

  • Add calls: plan version determined by the factual memory rule's plan_version. Defaults to Pro.
  • Search calls: plan version controlled by request parameter plan_version, independent of the rule. Defaults to Pro.
  • Updating a rule's plan_version affects new writes only; existing memories are unchanged.
  • Pre-commercialization rules default to Pro.

NoteInjecting retrieved memories into the prompt increases Token consumption. Fees are based on actual LLM Token usage.

Billing examples

ScenarioCalculationCost
Pro write 100 factual memory calls100 × ¥0.03¥3.00
Lite write 100 factual memory calls100 × ¥0.018¥1.80
Pro write 100 factual memory (multimodal) calls100 × ¥0.1¥10.00
Pro write 100 skill memory calls100 × ¥0.3¥30.00
Pro search 1,000 calls1,000 × ¥0.001¥1.00
Lite search 1,000 calls1,000 × ¥0.00002¥0.02
Store 10K entries for 1 month1 × ¥0.002 × 720 hours¥1.44

ImportantSee Rate limits and FAQ for more.

Previous：ReferenceNext：Rate Limits