> ## Documentation Index > Fetch the complete documentation index at: https://docs.layerrail.com/llms.txt > Use this file to discover all available pages before exploring further. # Pricing > LayerRail pricing for compute, storage, backups, Edge, monitoring, Game VPS, managed PostgreSQL, Kubernetes, networking, GitHub runners, and AI inference models. Updated on: July 24, 2026 LayerRail pricing is usage-based. A project is billed for the resources it creates, the time those resources run, and metered usage such as AI inference tokens, runner minutes, storage, cache, monitoring, and bandwidth. Prices below are the public pricing guide for the current LayerRail platform surface. Final checkout pages and monthly invoices are the source of truth for exact totals because region, storage, HA options, promotions, taxes, and usage can change the final bill. ## What LayerRail bills for | Product area | Billing model | Notes | | ------------------ | ----------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------ | | Virtual machines | Per running resource | Compute, storage, public IPv4, GPUs where selected, and related usage. | | Game VPS | Fixed plan pricing | Windows VPS plans billed by selected package. | | Managed PostgreSQL | Per running resource | vCPU, storage, high availability replicas, backups, and related infrastructure. | | Managed Kubernetes | Per cluster and worker capacity | Control plane, worker nodes, node storage, load balancers, and related usage. | | Object storage | Per provisioned bucket storage | S3-compatible bucket storage billed by stored GB over time. | | Block volumes | Per provisioned volume storage | Standalone block volumes billed by provisioned GB over time. | | Backups | Per backup storage usage | Backup storage billed by retained GB over time. | | Edge | Per provisioned Edge service | Edge services are billed while active. Bandwidth and request usage may be added as metering expands. | | Monitoring | Per backend and rule | Metrics storage, log storage, ingestion, uptime checks, and alert rules are billable. | | Networking | Per provisioned resource or usage | Load balancers, public IPv4, firewall-backed networking, and bandwidth where applicable. | | GitHub runners | Per runner minute, cache, and concurrency | Project-scoped runner compute used for CI workloads. | | AI inference | Per token or model operation | Premium Azure Foundry models are metered. Cloudflare Workers AI models are available through the LayerRail AI surface with project quota and usage tracking. | ## Billing rules * Duration-based resources are billed per minute and capped at `672` hours per monthly invoice window for the same resource slot. * Compute resources start billing when provisioning begins and stop billing when deletion is completed or the resource is marked deleted by the control plane. * Storage-backed resources bill by provisioned or retained GB over time until deleted. * Usage-based AI charges are recorded per project and attached to the project invoice. * Free AI quota is consumed before paid inference charges where eligible. * Premium AI usage is metered separately and can be charged through Polar usage metering. * GitHub runner projects receive a monthly `$1.00` runner credit when runner usage exists. * Outstanding invoices remain visible until paid. * If a project has no valid payment method and has unpaid invoices, services can be restricted, suspended, or deleted according to the billing terms. * Coupons and discounts are not applied to outstanding invoice settlement unless LayerRail explicitly issues a billing credit. ## Virtual machines Virtual machines are billed from the selected plan, region, storage, public IP, and optional GPU configuration. | Component | How it is priced | | --------------- | ---------------------------------------------------------------- | | vCPU and memory | Included in the VM plan shown during creation. | | Storage | Included or metered depending on the plan and storage size. | | Public IPv4 | Included where marked, otherwise billed as a public IP resource. | | GPU | Metered when a GPU plan is selected. | | Bandwidth | May be metered where applicable. | Use the VM create page in the console for exact regional plan totals before launch. ## Game VPS Game VPS plans are fixed packages for game communities and Windows-based server workloads. Every package includes the same 128 GB Windows system SSD. Higher tiers add CPU and RAM instead of increasing storage. | Plan | vCPU | RAM | Windows system SSD | Monthly price | | --------- | ---: | ----: | -----------------: | ------------: | | Starter | 2 | 4 GB | 128 GB | \$3.00/mo | | Community | 2 | 8 GB | 128 GB | \$5.00/mo | | Squad | 4 | 8 GB | 128 GB | \$8.00/mo | | Growth | 4 | 16 GB | 128 GB | \$16.00/mo | | Serious | 8 | 32 GB | 128 GB | \$40.00/mo | | Arena | 16 | 64 GB | 128 GB | \$75.00/mo | Game VPS checkout shows the exact payment amount before provisioning. ## Managed PostgreSQL Managed PostgreSQL pricing is based on the selected database flavor, vCPU size, storage, region, and availability mode. | Component | How it is priced | | ----------------- | --------------------------------------------------------------- | | Compute | vCPU-based pricing by selected database size. | | Storage | Metered by provisioned storage size. | | High availability | Additional standby capacity is billed when enabled. | | Backups | Backup storage and retention may affect usage. | | Networking | Public/private networking and firewall-related usage may apply. | PostgreSQL logs, backups, HA, and networking are part of the managed service experience. The create page and invoice show the exact selected size and usage. ## Managed Kubernetes Kubernetes billing includes the cluster control plane, worker nodes, node storage, and networking resources used by the cluster. | Component | How it is priced | | ------------------------ | -------------------------------------------------- | | Control plane | Billed for managed cluster control plane capacity. | | Worker nodes | Billed by selected worker size and node count. | | Node storage | Metered by provisioned worker storage. | | Load balancers | Billed when services expose public traffic. | | Public IPs and bandwidth | Metered where applicable. | Scaling worker nodes increases the project bill. Deleting a cluster stops new cluster billing after cleanup completes. ## Storage LayerRail now exposes S3-compatible object storage, standalone block volumes, and backup storage as billable platform primitives. Monthly equivalents below use the platform invoice cap of `672` hours per month. Exact invoices are calculated per minute. | Resource | Billing unit | Public guide price | | --------------------- | -------------------- | -----------------: | | Object bucket storage | Stored GB-month | \$0.0187/GB-mo | | Block volume storage | Provisioned GB-month | \$0.0373/GB-mo | | Backup storage | Retained GB-month | \$0.0373/GB-mo | ### Object storage Object storage is S3-compatible. Buckets are billed by storage used over time. Public bucket URLs and S3 endpoints do not change the storage rate. ### Block volumes Standalone block volumes are billed by provisioned size while the volume exists. Attaching a volume to a VM does not stop the volume storage charge. ### Backups Backups are billed by retained backup storage. Backup charges can apply to VMs, PostgreSQL, and other services where backup retention is enabled. ## Networking Networking resources can add cost when they reserve public capacity or process traffic. | Resource | Pricing behavior | | -------------- | ----------------------------------------------------------------------------------------- | | Load balancers | Billed while provisioned. | | Public IPv4 | Billed when not included in the selected plan. | | Firewalls | Included with networking controls unless attached usage requires billable infrastructure. | | Bandwidth | May be metered depending on product and region. | ## Edge LayerRail Edge services are billed while provisioned. | Resource | Billing unit | Public guide price | | --------------------- | -------------------- | -----------------: | | Standard Edge service | Active service-month | \$5.60/mo | Edge services currently cover origin routing, hostname mapping, and cache mode configuration. Request, bandwidth, and cache egress metering may appear separately as the Edge product expands. ## Monitoring LayerRail Monitoring is billed separately for metrics, logs, uptime checks, and alerting. Metrics and log backends are managed resources, while uptime checks and alert rules are billed as always-on monitoring controls. Monthly equivalents below use the platform invoice cap of `672` hours per month. Exact invoices are calculated per minute. | Resource | Billing unit | Public guide price | | ----------------- | ------------- | -----------------: | | Metrics storage | GB-month | \$0.0373/GB-mo | | Metrics ingestion | Backend-month | \$2.80/mo | | Log storage | GB-month | \$0.0373/GB-mo | | Log ingestion | Backend-month | \$2.80/mo | | Uptime check | Check-month | \$0.93/mo | | Alert rule | Rule-month | \$0.47/mo | ## GitHub runners LayerRail GitHub runners are billed by runner minutes, cache storage, and optional concurrency. Use them when you want CI workloads to run on project-managed infrastructure instead of shared hosted runners. Runner minutes are recorded after a GitHub job starts on a LayerRail runner. Standard and ARM standard runners share the same public minute pricing. | Runner label family | Price | | --------------------------------- | -----------: | | `standard-2` / `standard-2-arm` | \$0.0010/min | | `standard-4` / `standard-4-arm` | \$0.0020/min | | `standard-8` / `standard-8-arm` | \$0.0040/min | | `standard-16` / `standard-16-arm` | \$0.0080/min | | `standard-30` / `standard-30-arm` | \$0.0150/min | | `standard-60` / `standard-60-arm` | \$0.0300/min | | `premium-2` | \$0.0016/min | | `premium-4` | \$0.0032/min | | `premium-8` | \$0.0064/min | | `premium-16` | \$0.0120/min | | `premium-30` | \$0.0240/min | | Additional runner item | Billing unit | Public guide price | | ----------------------------- | ----------------- | -----------------: | | GitHub runner cache storage | GB-month | \$0.0400/GB-mo | | Additional runner concurrency | Concurrency-month | \$1.00/mo | Each project receives a monthly `$1.00` GitHub runner credit when runner usage is present. ## AI inference LayerRail AI has two provider classes: | Provider | Use case | Billing | | --------------------- | ------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------- | | Azure Foundry | Premium GPT, Claude, DeepSeek, Mistral, Grok, Llama, and Phi models deployed through LayerRail. | Metered by input and output tokens. | | Cloudflare Workers AI | Open and utility models for text, speech, embeddings, image, translation, and classification workloads. | Usage tracked through project quota and LayerRail billing policy. | ### Free quota Projects currently receive `100,000` free inference tokens. Free quota refreshes monthly and is consumed before eligible paid inference usage. ### Premium text model pricing Prices are shown per 1M tokens. | Model | Provider | Context | Input | Output | | ----------------- | ------------- | ------: | ------: | -------: | | GPT-5 | Azure Foundry | 400K | \$1.75 | \$14.00 | | GPT-5 Mini | Azure Foundry | 400K | \$0.35 | \$2.80 | | GPT-5 Nano | Azure Foundry | 400K | \$0.08 | \$0.60 | | GPT-5.5 | Azure Foundry | 400K | \$2.50 | \$20.00 | | GPT-5.4 | Azure Foundry | 400K | \$2.00 | \$16.00 | | GPT-5.4 Mini | Azure Foundry | 400K | \$0.45 | \$3.60 | | GPT-5.4 Nano | Azure Foundry | 400K | \$0.12 | \$0.90 | | GPT-5.2 | Azure Foundry | 400K | \$1.80 | \$14.40 | | GPT-4.1 | Azure Foundry | 1M | \$2.80 | \$11.20 | | GPT-4.1 Mini | Azure Foundry | 1M | \$0.55 | \$2.20 | | GPT-4o | Azure Foundry | 128K | \$3.50 | \$14.00 | | GPT-4o Mini | Azure Foundry | 128K | \$0.21 | \$0.84 | | o3 | Azure Foundry | 200K | \$2.80 | \$11.20 | | o4 Mini | Azure Foundry | 200K | \$0.55 | \$2.20 | | Claude Opus 4.8 | Azure Foundry | 1M | \$7.00 | \$35.00 | | Claude Fable 5 | Azure Foundry | 1M | \$21.00 | \$105.00 | | Claude Sonnet 5 | Azure Foundry | 1M | \$4.20 | \$21.00 | | Claude Sonnet 4.6 | Azure Foundry | 1M | \$4.20 | \$21.00 | | Claude Sonnet 4.5 | Azure Foundry | 1M | \$4.20 | \$21.00 | | Claude Haiku 4.5 | Azure Foundry | 200K | \$1.40 | \$7.00 | | Claude Opus 4.1 | Azure Foundry | 1M | \$21.00 | \$105.00 | | DeepSeek V4 Flash | Azure Foundry | 128K | \$0.40 | \$1.60 | | DeepSeek V4 Pro | Azure Foundry | 128K | \$0.80 | \$3.20 | | Mistral Large 3 | Azure Foundry | 128K | \$2.80 | \$8.40 | | Grok 4 | Azure Foundry | 256K | \$3.00 | \$15.00 | | Llama 4 Scout 17B | Azure Foundry | 128K | \$0.30 | \$0.90 | | Phi-4 Mini | Azure Foundry | 128K | \$0.10 | \$0.40 | ### Cloudflare Workers AI model catalog Cloudflare Workers AI models are exposed through the same LayerRail AI surface for eligible projects. | Capability | Available models | | ------------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Text generation | Llama 3.2 3B Instruct, GPT OSS 20B, GPT OSS 120B, Gemma SEA-LION v4 27B IT, Kimi K2.5, DeepSeek R1 Distill Qwen 32B, Gemma, Granite, Llama 2, Llama 3, Llama 3.1, Llama 3.2, Llama 3.3, Llama 4 Scout, Phi-2, SQLCoder, Mistral, Hermes, Kimi K2.6, Kimi K2.7 Code, NVIDIA Nemotron, Qwen, QwQ, GLM. | | Embeddings | BGE Base English v1.5, BGE Large English v1.5, BGE M3, BGE Small English v1.5, EmbeddingGemma 300M, PLaMo Embedding 1B, Qwen3 Embedding 0.6B. | | Speech to text | Flux, nova-3, Whisper, Whisper Large v3 Turbo, whisper-tiny-en. | | Text to speech | aura-1, Aura 2 English, Aura 2 Spanish, MeloTTS. | | Text to image | Flux 1 Schnell, Flux 2 Dev, Flux 2 Klein, SDXL Lightning, Lucid Origin, Phoenix 1.0, DreamShaper, Stable Diffusion. | | Vision and image understanding | LLaVA 1.5 7B, UForm Gen2 Qwen 500M, ResNet-50, DETR ResNet-50. | | Classification and reranking | BGE Reranker Base, DistilBERT SST-2. | | Summarization | BART Large CNN. | | Translation | IndicTrans2 EN to Indic 1B, M2M100 1.2B. | | Voice activity detection | Smart Turn v2. | ## Invoices and payment LayerRail invoices show the resources, usage period, unit price, usage amount, and total cost for the billing period. Previous invoices remain available in the billing page. If an invoice is unpaid, the project owner can pay it from the billing page. Outstanding invoice payments use fixed invoice amounts. ## Questions For pricing, invoices, credits, or payment issues, contact [support@layerrail.com](mailto:support@layerrail.com).