Kedis ku xidhan hagaha · Adag Heerka

Wadarta Kharashka Lahaanshaha ee Kediska LLM-yada is-martigeliyay

Tijaabi waxa keenaya qiimaha dhabta ah ee is-martigelinta LLMs: GPU xusuusta, ka faa'iidaysiga, shaqaaleysiinta iyo jebinta xitaa ka soo horjeeda qiimaha API-kiiba.

Wadooyinka haga ee la xidhiidhaWadarta Qiimaha Lahaanshaha LLM-yada is-xilqaamay
Su'aal 1 ee 8

Qiyaastii intee in le'eg xusuusta ayaa culeyska moodelka 70-bilyan ee cabbirka u baahan yahay 16-bit oo sax ah?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Qiyaastii intee in le'eg xusuusta ayaa culeyska moodelka 70-bilyan ee cabbirka u baahan yahay 16-bit oo sax ah?
    • About 140 GB
    • About 35 GB
    • About 70 GB
    • About 560 GB
  2. Sidee buu hagaygu u qeexaya qiimaha wax ku oolka ah ee calaamad kasta ee geynta is-xilqaanka ah?
    • GPU hourly rental price divided by the model's parameter count
    • The API provider's list price minus a fixed discount
    • Total monthly cost divided by tokens actually served
    • Hardware purchase price divided by the number of users
  3. Waa maxay sababta koox si fudud loo isticmaalo oo is-marti galisay kharash ka badan halkii calaamad?
    • Self-hosted models produce longer answers
    • GPUs cost the same whether they are idle or busy, so fewer tokens share the fixed cost
    • Open-weight licenses charge a fee per token
    • APIs are subsidized by governments
  4. Kharashkee ayuu hagahu tilmaamayaa inta badan khadka qarsoon ee ugu weyn ee is-martigelinta?
    • Model download fees
    • Tokenizer licenses
    • Electricity for laptops
    • Staff who deploy, tune, patch, upgrade and stay on call for the stack
  5. Miisaanka moodeelka ka sokow, maxaa isticmaala xusuusta GPU-da oo ku koraa dhererka macnaha guud iyo tirada codsiyada isla socda?
    • Erayada tokenizer
    • The load balancer
    • The KV cache
    • The model license file
  6. Why is it a misconception that self-hosting is automatically more private?
    • A poorly secured self-hosted server can be riskier than a provider with audited controls
    • Open-weight models send data back to their creators
    • APIs never store any data
    • Privacy laws ban self-hosting
  7. What is the purpose of continuous batching in servers such as vLLM?
    • To compress model weights to 4 bits
    • To translate prompts into other languages
    • To download new model versions automatically
    • To keep GPUs busy by processing many requests together
  8. What does the guide describe as the likely pattern for many organizations?
    • Moving every workload to self-hosting
    • Hybrid: APIs for spiky or frontier workloads, self-hosting for steady, high-volume or sensitive ones
    • Using only APIs forever
    • Choosing once and never revisiting the decision