Closed-FT vs Self-Hosted FT Decision Matrix: TCO + Latency + Data Residency + KVKK
Cookbook's Part XIV summary decision: closed API FT vs self-hosted open FT. 6-dim comparison: TCO (1-yr estimate), latency (P50/P95), data residency (TR/EU/US), KVKK compliance, model freedom (versioning, license, deploy), quality. Typical decisions for 4 use-cases.
Şükrü Yusuf KAYA
26 min read
Intermediate
1. 6-Boyut Karar Matrisi#
| Boyut | Closed API FT | Self-Hosted FT |
|---|---|---|
| TCO (1 yıl) | OPEX (saatlik) | CAPEX (rig) + OPEX (elektrik) |
| Latency | API overhead 100-300ms | local 50ms |
| Data residency | OpenAI US, Anthropic US/EU, AWS multi-region | tam kontrol |
| KVKK | OpenAI ToS okuma + DPA | tam compliant |
| Model freedom | tied to provider | tam özgür |
| Kalite | GPT-5/Claude-3.7 SOTA | Llama 3.3 70B en yakın |
2. 4 Use-Case Karar Tablosu#
| Use case | Cookbook tavsiye |
|---|---|
| Startup MVP, 1000 user/gün | Closed (OpenAI GPT-4o-mini FT) |
| TR finans/bankacılık (KVKK kritik) | Self-hosted (Qwen 2.5 7B + on-prem) |
| 100K user/gün, high traffic | Self-hosted (cost ekonomi) |
| Research / paper | Self-hosted (reproducibility) |
| Quick experimentation | Closed (no infra setup) |
| TR kamu/devlet | Self-hosted (yerel veri) |
✅ Part XIV tamamlandı
- Kendi use-case'in için karar matrisini doldur. 2) Sonraki Part: Part XVI — Production Operations.
Yorumlar & Soru-Cevap
(0)Yorum yazmak için giriş yap.
Yorumlar yükleniyor...
Related Content
Part 0 — Engineering Foundations
Welcome to the Fine-Tuning Cookbook: System, Stage Taxonomy, and the Reproducibility Contract
Start LearningPart 0 — Engineering Foundations
Reproducibility Stack: Seeds, cuDNN Flags, and Deterministic CUDA — End the 'Works on My Machine' Problem
Start LearningPart 0 — Engineering Foundations