macrostack
Browse

The AI stack

Categories

Local & Sovereign AINotes & KnowledgeObservability & MonitoringPassword ManagersWeb AnalyticsTeam ChatSmart HomeNetworking & RoutersVideo ConferencingCloud Storage & SyncPhotos & MediaAPI DevelopmentImage EditingWorkflow Automation & iPaaSDeveloper Tools & ContainersOffice & Productivity SuitesNo-Code DatabasesCode Hosting & Git ForgesProject ManagementEmail Marketing & NewslettersScheduling & BookingError Tracking & Exception MonitoringLog Management & SIEMVPN & PrivacyEmail & Secure MailVector Databases & AI SearchLLM & Agent FrameworksDomains & Web HostingData Removal & PrivacyAuthentication & IdentityHelp Desk & Customer SupportCloud & VPSKubernetes & Container PlatformsEmbedding ModelsPDF & DocumentsAI Coding AssistantsAI Voice & SpeechLLM Observability & EvaluationLLM Gateways & RoutingCloud GPU & AI ComputeCI/CD & build automationData & pipeline orchestrationModel serving & inferenceAI agent frameworksBackend as a serviceSecrets managementFeature flags & experimentationProduct analyticsSearch infrastructureUptime & status monitoringAffiliate & partner platformsVisitor identification & personalisationWikis & internal docsIdentity & access managementData warehouses & analytics enginesCustomer data platformsCRMObject storageBI & dashboardsE-signatureWhiteboards & diagrammingIn-memory data stores & cachingPlatform as a serviceTransactional & bulk emailHeadless CMSDesign & prototypingE-commerce platformsInternal tools & admin panelsManaged databasesForms & surveysFine-Tuning & Model TrainingRAG & Retrieval PlatformsLLM Evaluation & TestingAI Guardrails & Content SafetySpeech Recognition & TranscriptionExperiment Tracking & ML OpsDocument AI & OCR

About

How we rank & score
Migration guide · Layer 4

OpenAI API (ChatGPT) vLLM

What it saves, what actually moves, what you rebuild — and the thing that catches people.

What it costs, and what it saves

At scale the comparison flips: an H100 at roughly $2/hr serving continuously costs about $1,400/month and handles throughput that would cost far more per token via API. Below heavy sustained use, the API is cheaper.

88
vLLM
Apache-2.0
Effort: A week to serve. Ongoing to operate.

Moves cleanly

Application code, via vLLM's OpenAI-compatible server. Same base-URL change as Ollama.

You rebuild

Prompts, plus the serving infrastructure Ollama would have hidden from you.

What OpenAI API (ChatGPT) costs you today

Usage-based per-token billing; ChatGPT Plus/Team/Enterprise seats on top

What actually holds you in

Proprietary models and API surface; your data and cost structure sit entirely on OpenAI's terms.

What you are moving to

vLLM is a fast inference and serving engine built for throughput, using paged attention to serve many concurrent requests efficiently. It is the choice when a team needs to self-host models at real scale.

Free / self-host

vLLM strengths

  • Excellent throughput under concurrency
  • OpenAI-compatible server mode
  • Backed by a large community

What you give up

  • Aimed at capable GPUs, not laptops
  • Steeper operational learning curve

The migration, step by step

  1. 1Benchmark honestly: measure your actual tokens per day before assuming self-hosting is cheaper
  2. 2Rent a GPU and serve with vllm serve <model> --api-key <key>
  3. 3Point your OpenAI client at it
  4. 4Tune tensor parallelism and max_model_len for your GPU; defaults leave throughput on the table
  5. 5Add health checks and a supervisor, because nothing restarts it for you

The gotcha

vLLM is a throughput engine, not a convenience layer. It rewards understanding PagedAttention and batching and punishes treating it as a black box — the same GPU can differ severalfold in throughput on configuration alone.

When to stay on OpenAI API (ChatGPT)

Your volume is bursty or modest — a GPU idles at full price while an API bills zero.

Compare firstAll 8 OpenAI API (ChatGPT) alternatives, rankedWhere this sitsLayer 4Models & tooling

Other ways off OpenAI API (ChatGPT)

More Layer 4 migrations

Figures verified 2026-07-04 against vendor pricing pages. Prices change and migrations differ by estate — treat the cost delta as a starting model, not a quote. Rankings and recommendations here are merit-only; affiliate income never changes a verdict. See our methodology.

The Macrostack brief

New swaps, worth your inbox.

A short, occasional email when we add a high-intent alternative or ship a new head-to-head. No spam, no selling your address — unsubscribe in one click.