{
 "features_resolved": {
  "snapshot_any": true,
  "snapshot_mem": false,
  "fork": false,
  "pause_resume": true,
  "persistent_disk": true,
  "whole_disk": false,
  "persist_scope": true,
  "volumes": true,
  "auto_stop_idle": false,
  "long_sessions": true,
  "custom_image": true,
  "image_snapshot": true,
  "image_oci": true,
  "vm_isolation": false,
  "docker_inside": false,
  "nested_virt": false,
  "root": true,
  "systemd": false,
  "browser": false,
  "desktop": false,
  "computer_use": false,
  "code_interpreter": true,
  "browser_control": null,
  "desktop_control": null,
  "browser_and_desktop": false,
  "anti_bot": false,
  "captcha_solving": false,
  "residential_ip": false,
  "agent_harnesses": false,
  "ssh": true,
  "public_ipv4": true,
  "inbound_https": true,
  "custom_domain": null,
  "raw_tcp_inbound": true,
  "egress_allowlist": false,
  "open_internet": true,
  "static_egress_ip": false,
  "private_network": true,
  "self_host": false,
  "byoc": false,
  "open_source": false,
  "soc2": true,
  "hipaa": true,
  "sso": true,
  "gpu": true,
  "arm64": false,
  "windows": false,
  "macos": false,
  "lang_python": true,
  "lang_node": true,
  "lang_go": null,
  "lang_rust": null,
  "lang_java": null,
  "gpu_desktop": false,
  "credential_injection": null,
  "egress_http_rules": false,
  "wake_on_request": true,
  "live_resize": false,
  "fork_running": false,
  "memory_snapshot_fork": false,
  "webhooks": true,
  "scheduled_wakeups": null,
  "audit_logs": true,
  "eu_data_residency": true,
  "zero_data_retention": null,
  "scoped_api_keys": true,
  "spend_limits": true,
  "mcp_server": true,
  "usage_api": false,
  "snapshot_auto": false,
  "snapshot_on_demand": false,
  "harness_api": false,
  "own_agent_api": false,
  "fast_boot": null,
  "strong_isolation": false,
  "ingress_rules": true,
  "guest_firewall": false,
  "secret_proxy": false,
  "secret_proxy_any": false,
  "volume_attach": true,
  "volume_shared": true,
  "sidecar": null,
  "fuse": null,
  "kernel_features": null,
  "burst": null
 },
 "company": {
  "id": "runpod",
  "checked": "2026-09-30",
  "company": "Runpod, Inc.",
  "ownership": {
   "value": "independent",
   "parent": null,
   "source": "https://techcrunch.com/2026/01/16/ai-cloud-startup-runpod-hits-120m-in-arr-and-it-started-with-a-reddit-post/",
   "quote": "Runpod, an AI app hosting platform that launched four years ago, has hit a $120 million annual revenue run rate, founders Zhen Lu and Pardeep Singh tell TechCrunch."
  },
  "stage": {
   "value": "series-a",
   "source": "https://siliconangle.com/2026/06/25/runpod-raises-100m-build-leading-cloud-platform-ai-developers/",
   "quote": "The Series A funding, led by Summit Partners, brings the company's total funding to $122 million across multiple rounds, including a seed round in May 2024 co-led by Intel Capital and Dell Technologies Capital.",
   "confidence": "high",
   "note": "Bootstrapped 2021-2024 before any outside capital; first priced round was a $20M seed (May 2024), followed directly by a $100M Series A (June 2026) — no intermediate seed extension or bridge reported."
  },
  "accelerators": [
   {
    "name": "AI Grant",
    "batch": "2",
    "source": "https://aigrant.com/",
    "quote": "listed in the AI Grant portfolio as \"RunPod\"",
    "confidence": "high",
    "via": "portfolio-domain"
   }
  ],
  "total_funding_usd": {
   "value": 122000000,
   "status": "confirmed",
   "basis": "SiliconANGLE reporting on the Series A announcement states total funding of $122M across the seed and Series A rounds",
   "source": "https://siliconangle.com/2026/06/25/runpod-raises-100m-build-leading-cloud-platform-ai-developers/",
   "quote": "brings the company's total funding to $122 million across multiple rounds"
  },
  "last_round": {
   "type": "series-a",
   "amount_usd": 100000000,
   "date": "2026-06",
   "leads": [
    "Summit Partners"
   ],
   "status": "confirmed",
   "source": "https://siliconangle.com/2026/06/25/runpod-raises-100m-build-leading-cloud-platform-ai-developers/",
   "quote": "Artificial intelligence developer-centric cloud provider Runpod Inc. announced Wednesday that it has raised $100 million in early-stage funding, pushing the valuation of the company to $1 billion. The Series A funding, led by Summit Partners..."
  },
  "investors": [
   "Summit Partners",
   "Dell Technologies Capital",
   "Intel Capital",
   "Nat Friedman",
   "Julien Chaumond"
  ],
  "portfolio_listings": [
   {
    "source": "AI Grant",
    "company": "RunPod",
    "batch": "2",
    "how": "domain"
   }
  ],
  "revenue": {
   "value_usd": 120000000,
   "kind": "ARR",
   "as_of": "2026-01",
   "status": "confirmed",
   "basis": "Company press release; corroborated by TechCrunch",
   "source": "https://www.runpod.io/press/runpod-ai-cloud-surpasses-120m-in-arr",
   "quote": "Runpod ... today announced it has surpassed $120 million in annual recurring revenue (ARR) ... revenue grew 90% YoY"
  },
  "employees": {
   "value": null,
   "source": ""
  },
  "note": "Founded late 2021/early 2022 by Zhen Lu and Pardeep Singh (ex-Comcast developers) after repurposing Ethereum-mining rigs into GPU servers; grew via Reddit posts and revenue-share data-center deals, bootstrapping to ~$24M revenue before any VC money. Took a $20M seed co-led by Dell Technologies Capital and Intel Capital in May 2024 (angels incl. Nat Friedman and Hugging Face co-founder Julien Chaumond), then a $100M Series A led by Summit Partners in June 2026 at a $1B valuation. Surpassed $120M ARR (Jan 2026) serving 1M+ developers; customers include Replit, Cursor, OpenAI, Perplexity, Wix, Zillow."
 },
 "id": "runpod",
 "name": "Runpod",
 "url": "https://www.runpod.io/pricing",
 "category": "gpu-cloud",
 "isolation": "container",
 "modes": [
  {
   "key": "cpu-pod",
   "label": "CPU Pod, Secure Cloud, compute-optimized (cpu3c)",
   "flags": [],
   "os": [
    "linux"
   ],
   "regions": [
    "us",
    "eu",
    "asia",
    "other"
   ],
   "gpu_only": false,
   "features": {},
   "pricing": "sizes",
   "cpu_basis": "alloc",
   "ram_basis": "alloc",
   "active_floor": 0,
   "sizes": [
    {
     "name": "cpu3c-2-4",
     "vcpu": 2,
     "ram_gib": 4,
     "disk_gib": null,
     "hour": 0.06,
     "month_cap": null
    },
    {
     "name": "cpu3c-4-8",
     "vcpu": 4,
     "ram_gib": 8,
     "disk_gib": null,
     "hour": 0.12,
     "month_cap": null
    },
    {
     "name": "cpu3c-8-16",
     "vcpu": 8,
     "ram_gib": 16,
     "disk_gib": null,
     "hour": 0.24,
     "month_cap": null
    },
    {
     "name": "cpu3c-16-32",
     "vcpu": 16,
     "ram_gib": 32,
     "disk_gib": null,
     "hour": 0.48,
     "month_cap": null
    },
    {
     "name": "cpu3c-32-64",
     "vcpu": 32,
     "ram_gib": 64,
     "disk_gib": null,
     "hour": 0.96,
     "month_cap": null
    },
    {
     "name": "cpu3g-2-8",
     "vcpu": 2,
     "ram_gib": 8,
     "disk_gib": null,
     "hour": 0.08,
     "month_cap": null
    },
    {
     "name": "cpu3g-4-16",
     "vcpu": 4,
     "ram_gib": 16,
     "disk_gib": null,
     "hour": 0.16,
     "month_cap": null
    },
    {
     "name": "cpu3g-8-32",
     "vcpu": 8,
     "ram_gib": 32,
     "disk_gib": null,
     "hour": 0.32,
     "month_cap": null
    },
    {
     "name": "cpu3g-16-64",
     "vcpu": 16,
     "ram_gib": 64,
     "disk_gib": null,
     "hour": 0.64,
     "month_cap": null
    },
    {
     "name": "cpu3g-32-128",
     "vcpu": 32,
     "ram_gib": 128,
     "disk_gib": null,
     "hour": 1.28,
     "month_cap": null
    },
    {
     "name": "cpu3m-2-16",
     "vcpu": 2,
     "ram_gib": 16,
     "disk_gib": null,
     "hour": 0.11,
     "month_cap": null
    },
    {
     "name": "cpu3m-4-32",
     "vcpu": 4,
     "ram_gib": 32,
     "disk_gib": null,
     "hour": 0.22,
     "month_cap": null
    },
    {
     "name": "cpu3m-8-64",
     "vcpu": 8,
     "ram_gib": 64,
     "disk_gib": null,
     "hour": 0.44,
     "month_cap": null
    },
    {
     "name": "cpu3m-16-128",
     "vcpu": 16,
     "ram_gib": 128,
     "disk_gib": null,
     "hour": 0.88,
     "month_cap": null
    },
    {
     "name": "cpu3m-32-256",
     "vcpu": 32,
     "ram_gib": 256,
     "disk_gib": null,
     "hour": 1.76,
     "month_cap": null
    }
   ],
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": 0,
   "granularity_s": 1,
   "start_fee": 0,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": null,
   "note": "cpu3c Compute-Optimized secure pod price from Runpod's public GraphQL catalogue (api.runpod.io/graphql, no key needed, 2026-09-29): securePrice $0.03/vCPU-h with RAM bundled at 2 GB/vCPU (ramMultiplier 2), 2-32 vCPU, container disk limit 10 GB/vCPU. The docs API v2 example ($0.04, 2.5 GB/vCPU) is illustrative only. Other flavours (same API): cpu3g General Purpose 4 GB/vCPU $0.04/vCPU-h; cpu3m Memory-Optimized 8 GB/vCPU $0.055; cpu5c 2 GB/vCPU $0.035; cpu5g 4 GB/vCPU $0.046; cpu5m 8 GB/vCPU $0.065. Billed per second while running; container/volume disk extra. Deploy requires >= 1 h of credit for the config. Allocation billing: \"Pods are billed by the second for compute and storage\" (docs.runpod.io/pods/pricing); no idle auto-stop. Provisioning time unpublished (boot_overhead_s null). GraphQL API is announced for retirement in early 2027 (release notes)."
  },
  {
   "key": "gpu-secure",
   "label": "GPU Pod, Secure Cloud on-demand (per GPU, vCPU/RAM bundled)",
   "flags": [],
   "os": [
    "linux"
   ],
   "regions": [
    "us",
    "eu",
    "asia",
    "other"
   ],
   "gpu_only": true,
   "features": {},
   "pricing": "resource",
   "vcpu_h": 0,
   "ram_gib_h": 0,
   "cpu_basis": "alloc",
   "ram_basis": "alloc",
   "active_floor": 0,
   "ram_per_vcpu": [
    null,
    null
   ],
   "ram_fixed_per_vcpu": null,
   "vcpu_options": null,
   "min_vcpu": null,
   "max_vcpu": 32,
   "max_ram_gib": 283,
   "sizes": null,
   "pool_tiers": null,
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": 0,
   "granularity_s": 1,
   "start_fee": 0,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": {
    "B300": 8.99,
    "B200": 7.99,
    "H200": 5.29,
    "H100": 3.99,
    "H100-NVL": 3.19,
    "H100-PCIe": 2.89,
    "RTX-PRO-6000": 2.49,
    "A100-80GB": 1.79,
    "A100-80GB-SXM": 1.79,
    "L40S": 1.09,
    "RTX-PRO-6000-MIG-48GB": 1.29,
    "RTX-6000-Ada": 0.99,
    "L40": 0.82,
    "RTX-A6000": 0.59,
    "A40": 0.59,
    "RTX-5090": 1.19,
    "RTX-4090": 0.89,
    "RTX-PRO-6000-MIG-24GB": 0.69,
    "RTX-3090": 0.5,
    "L4": 0.59,
    "RTX-A5000": 0.27,
    "MI300X": 2.39,
    "H200-NVL": 3.79,
    "RTX-PRO-6000-WK": 2.19,
    "RTX-PRO-5000": 0.96,
    "RTX-PRO-4500": 0.72,
    "RTX-PRO-4000": 0.57,
    "RTX-4000-Ada": 0.28,
    "RTX-A4000": 0.25,
    "RTX-A4500": 0.25,
    "RTX-2000-Ada": 0.24
   },
   "note": "Per-GPU price includes a fixed per-GPU vCPU/RAM shape (e.g. H100 SXM 20 vCPU/125 GB, B200 28 vCPU/283 GB, RTX 4090 6 vCPU/41 GB, A40 9 vCPU/50 GB), so vcpu_h/ram_gib_h = 0. H100 = H100 SXM. Secure Cloud = T3/T4 data centers; the only option for Organizations/post-paid accounts. Prices from the runpod.io/pricing Pods grid with its default Secure Cloud toggle (\"GPU Cloud Pricing Updated October 8, 2026\"; rows carry data-secure, e.g. H100 SXM \"80 GB VRAM 125 GB RAM 20 vCPUs $ 3.99 /hr\"), checked 2026-10-14. The Oct 8 update raised 15 Secure rates (H100 SXM 3.49->3.99, B200 6.79->7.99, B300 7.89->8.99, H200 4.59->5.29, RTX 4090 0.74->0.89, A40 0.49->0.59 ...); H100 PCIe/NVL, L40S, L40, RTX 3090, RTX A5000 unchanged. Types not in the grid (MI300X, H200-NVL, RTX-PRO-6000-WK, RTX-PRO-5000/4500/4000, RTX-4000-Ada, RTX-A4000/A4500, RTX-2000-Ada) keep their 2026-09 catalogue rates and may also have moved."
  },
  {
   "key": "gpu-community",
   "label": "GPU Pod, Community Cloud on-demand (peer-to-peer hosts, variable reliability)",
   "flags": [
    "stock"
   ],
   "os": [
    "linux"
   ],
   "regions": [
    "us",
    "eu",
    "asia",
    "other"
   ],
   "gpu_only": true,
   "features": {},
   "pricing": "resource",
   "vcpu_h": 0,
   "ram_gib_h": 0,
   "cpu_basis": "alloc",
   "ram_basis": "alloc",
   "active_floor": 0,
   "ram_per_vcpu": [
    null,
    null
   ],
   "ram_fixed_per_vcpu": null,
   "vcpu_options": null,
   "min_vcpu": null,
   "max_vcpu": 32,
   "max_ram_gib": 283,
   "sizes": null,
   "pool_tiers": null,
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": 0,
   "granularity_s": 1,
   "start_fee": 0,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": {
    "B300": 6.94,
    "B200": 5.98,
    "H200": 3.59,
    "H100": 2.69,
    "H100-NVL": 2.59,
    "H100-PCIe": 1.99,
    "RTX-PRO-6000": 1.69,
    "A100-80GB": 1.19,
    "A100-80GB-SXM": 1.39,
    "L40S": 0.79,
    "RTX-6000-Ada": 0.74,
    "L40": 0.69,
    "RTX-A6000": 0.33,
    "RTX-5090": 0.69,
    "RTX-4090": 0.34,
    "RTX-3090": 0.22,
    "RTX-A5000": 0.16,
    "A100-40GB": 1,
    "RTX-PRO-6000-WK": 1.69,
    "RTX-PRO-5000": 0.82,
    "RTX-5080": 0.39,
    "RTX-4000-Ada": 0.2,
    "RTX-A4500": 0.19,
    "RTX-A4000": 0.17,
    "V100": 0.19,
    "V100-SXM2": 0.23
   },
   "note": "Community Cloud per-GPU prices from the schema.org Offer data embedded in runpod.io/pricing (2026-09-28). Peer-to-peer hosts with variable reliability; Runpod no longer accepts new Community hosts (existing capacity remains). Not usable by Organizations (post-paid). Not interruptible: on-demand, just cheaper hosts. MIG variants have no Community price listed. Runpod docs (pods/choose-a-pod) describe Community Cloud as \"Peer-to-peer providers\" with \"Variable\" reliability vs Secure Cloud \"T3/T4 data centers\", \"High redundancy\", \"Best for Production\". No SLA stated for Community.",
   "stock_note": "Community Cloud is shrinking (no new hosts); on 2026-09-29 only 8 GPU types had Community stock, all Low; H100/B200/B300/RTX 4090/5090 had none."
  },
  {
   "key": "serverless-flex",
   "label": "Serverless flex worker (GPU, scale to zero)",
   "flags": [
    "alt"
   ],
   "os": [
    "linux"
   ],
   "regions": [
    "us",
    "eu",
    "asia",
    "other"
   ],
   "gpu_only": true,
   "features": {
    "ssh": false,
    "persistent_disk": false,
    "pause_resume": false,
    "wake_on_request": true
   },
   "pricing": "resource",
   "vcpu_h": 0,
   "ram_gib_h": 0,
   "cpu_basis": "alloc",
   "ram_basis": "alloc",
   "active_floor": 0,
   "ram_per_vcpu": [
    null,
    null
   ],
   "ram_fixed_per_vcpu": null,
   "vcpu_options": null,
   "min_vcpu": null,
   "max_vcpu": null,
   "max_ram_gib": null,
   "sizes": null,
   "pool_tiers": null,
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": 0,
   "granularity_s": 1,
   "start_fee": 0,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": {
    "B300": 9.98,
    "B200": 8.64,
    "H200": 5.93,
    "RTX-PRO-6000": 3.49,
    "H100": 4.79,
    "A100-80GB": 2.72,
    "L40S": 1.75,
    "L40": 1.75,
    "RTX-6000-Ada": 1.75,
    "A40": 1.22,
    "RTX-A6000": 1.22,
    "RTX-5090": 1.58,
    "RTX-PRO-4500": 1.15,
    "RTX-4090": 1.1,
    "L4": 0.69,
    "RTX-A5000": 0.69,
    "RTX-3090": 0.69,
    "16GB-class": 0.58
   },
   "note": "Request/queue workers, not an interactive box. Billed per second (rounded up) from worker start to full stop: cold start + execution + idle timeout (default 5 s). Prices are per GPU class (e.g. '48GB PRO' = L40/L40S/6000 Ada/MIG 48GB; '24GB' = L4/A5000/3090/MIG 24GB; 16GB = A4000/A4500/RTX 4000/RTX 2000). Container disk ~$0.10/GB-month in 5-min intervals. Active (always-on) workers: discount via sales only, rate unpublished. The H100 key is the pricing page's \"80GB\" Serverless tier listed as \"H100\" at $4.79/h (Flex); the page does not say which H100 variant (SXM/PCIe/NVL)."
  },
  {
   "key": "serverless-cpu",
   "label": "Serverless CPU worker (cpu3c, API example rate)",
   "flags": [
    "alt"
   ],
   "os": [
    "linux"
   ],
   "regions": [
    "us",
    "eu",
    "asia",
    "other"
   ],
   "gpu_only": false,
   "features": {
    "ssh": false,
    "persistent_disk": false,
    "pause_resume": false,
    "wake_on_request": true
   },
   "pricing": "sizes",
   "cpu_basis": "alloc",
   "ram_basis": "alloc",
   "active_floor": 0,
   "sizes": [
    {
     "name": "cpu3c-1-2",
     "vcpu": 1,
     "ram_gib": 2,
     "disk_gib": null,
     "hour": 0.036,
     "month_cap": null
    },
    {
     "name": "cpu3g-1-4",
     "vcpu": 1,
     "ram_gib": 4,
     "disk_gib": null,
     "hour": 0.05,
     "month_cap": null
    },
    {
     "name": "cpu3c-2-4",
     "vcpu": 2,
     "ram_gib": 4,
     "disk_gib": null,
     "hour": 0.072,
     "month_cap": null
    },
    {
     "name": "cpu3g-2-8",
     "vcpu": 2,
     "ram_gib": 8,
     "disk_gib": null,
     "hour": 0.1,
     "month_cap": null
    },
    {
     "name": "cpu3c-4-8",
     "vcpu": 4,
     "ram_gib": 8,
     "disk_gib": null,
     "hour": 0.144,
     "month_cap": null
    },
    {
     "name": "cpu3g-4-16",
     "vcpu": 4,
     "ram_gib": 16,
     "disk_gib": null,
     "hour": 0.2,
     "month_cap": null
    },
    {
     "name": "cpu3c-8-16",
     "vcpu": 8,
     "ram_gib": 16,
     "disk_gib": null,
     "hour": 0.288,
     "month_cap": null
    },
    {
     "name": "cpu3g-8-32",
     "vcpu": 8,
     "ram_gib": 32,
     "disk_gib": null,
     "hour": 0.4,
     "month_cap": null
    }
   ],
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": 0,
   "granularity_s": 1,
   "start_fee": 0,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": null,
   "note": "Serverless CPU (and Flash CPU) price from Runpod's public GraphQL catalogue (slsPrice, 2026-09-29): cpu3c $0.036/vCPU-h (2 GB/vCPU), cpu5c $0.042, cpu3g $0.05 (4 GB/vCPU), cpu5g $0.055; memory-optimized cpu3m/cpu5m have slsPrice 0 (not offered on Serverless). Flash instance IDs go down to 1 vCPU (cpu3c-1-2 ... cpu3c-8-16). Max container disk 10 GB x vCPU for cpu3c/cpu3g, 15 GB x vCPU for cpu5c. Scale to zero; billed from worker start to full stop, rounded up to the second, incl. idle timeout (default 5 s). Max workers per account: 5 by default, rising with balance to 60 at $900+."
  },
  {
   "key": "instant-cluster",
   "label": "Instant Cluster (multi-node, per GPU, no commitment)",
   "flags": [],
   "os": [
    "linux"
   ],
   "regions": null,
   "gpu_only": true,
   "features": {},
   "pricing": "resource",
   "vcpu_h": 0,
   "ram_gib_h": 0,
   "cpu_basis": "alloc",
   "ram_basis": "alloc",
   "active_floor": 0,
   "ram_per_vcpu": [
    null,
    null
   ],
   "ram_fixed_per_vcpu": null,
   "vcpu_options": null,
   "min_vcpu": null,
   "max_vcpu": null,
   "max_ram_gib": null,
   "sizes": null,
   "pool_tiers": null,
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": 0,
   "granularity_s": 1,
   "start_fee": 0,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": {
    "H200": 4.31,
    "A100-80GB-SXM": 1.79
   },
   "note": "Multi-node only: docs.runpod.io/instant-clusters lists every GPU (B200/H200/H100/A100) as \"2-8 nodes (16-64 GPUs)\", i.e. 8 GPUs per node and a 2-node minimum, so gpu_min_count 16. Beyond 8 nodes (up to 512 GPUs) via sales. Pricing page per GPU-h: H200 SXM $4.31, A100 SXM $1.79; L40S/H100 SXM/B200 contact sales. Reserved Clusters (1-12+ month terms) are sales-only.",
   "gpu_min_count": 16
  }
 ],
 "storage": {
  "included_disk_gib": 0,
  "disk_gib_month": 0.1,
  "disk_billed_when_stopped": true,
  "snapshot_gib_month": 0.2,
  "paused_ram_billed": false
 },
 "network": {
  "egress_free_gib": null,
  "egress_gib": 0,
  "ipv4_month": null,
  "ipv4_available": true,
  "internet": {
   "default": "full"
  }
 },
 "plans": [
  {
   "name": "Pay-as-you-go (prepaid credits)",
   "fee": 0,
   "flags": [],
   "fee_is_credit": false,
   "included_usd": 0,
   "per_seat": 0,
   "concurrency": null,
   "max_session_h": null,
   "max_vcpu": null,
   "max_ram_gib": null,
   "sizes_allowed": null,
   "note": "Prepaid balance (min ~$10 to start; prepaid cards >= $100). Pods auto-stop at $0 balance (pods without a network volume are terminated, data lost). Default spend limit $80/hour across all resources (raised over time / via support). Deploy needs >= 1 h of credit for the config."
  },
  {
   "name": "Enterprise / Organization (post-paid, contracted rates)",
   "fee": null,
   "flags": [
    "sales"
   ],
   "fee_is_credit": null,
   "included_usd": null,
   "per_seat": null,
   "concurrency": null,
   "max_session_h": null,
   "max_vcpu": null,
   "max_ram_gib": null,
   "sizes_allowed": null,
   "note": "Invoiced in arrears at contracted rates; Secure Cloud only; business invoicing for >$5,000 transactions. Reserved clusters and active-worker discounts negotiated."
  }
 ],
 "free": {
  "monthly_credit": 0,
  "one_time_credit": 0
 },
 "features": {
  "linux": true,
  "windows": false,
  "macos": false,
  "arm64": false,
  "gpu": true,
  "gpu_types": [
   "B300",
   "B200",
   "H200",
   "H100",
   "H100-NVL",
   "H100-PCIe",
   "RTX-PRO-6000",
   "A100-80GB",
   "A100-80GB-SXM",
   "L40S",
   "L40",
   "RTX-6000-Ada",
   "RTX-A6000",
   "A40",
   "RTX-5090",
   "RTX-4090",
   "RTX-3090",
   "L4",
   "RTX-A5000",
   "RTX-PRO-6000-MIG-48GB",
   "RTX-PRO-6000-MIG-24GB",
   "MI300X",
   "H200-NVL",
   "RTX-PRO-6000-WK",
   "RTX-PRO-5000",
   "RTX-PRO-4500",
   "RTX-PRO-4000",
   "RTX-4000-Ada",
   "RTX-A4000",
   "RTX-A4500",
   "RTX-2000-Ada",
   "A100-40GB",
   "RTX-5080",
   "V100",
   "V100-SXM2"
  ],
  "snapshot": "fs",
  "pause_resume": true,
  "fork": false,
  "persistent_disk": true,
  "volumes": true,
  "auto_stop_idle": false,
  "max_session_h": null,
  "custom_image": true,
  "docker_inside": false,
  "nested_virt": false,
  "root": true,
  "systemd": false,
  "browser": false,
  "desktop": false,
  "computer_use": false,
  "code_interpreter": true,
  "agent_harnesses": false,
  "ssh": true,
  "public_ipv4": true,
  "inbound_https": true,
  "custom_domain": null,
  "raw_tcp_inbound": true,
  "egress_allowlist": false,
  "static_egress_ip": false,
  "private_network": true,
  "self_host": false,
  "byoc": false,
  "open_source": false,
  "regions": [
   "us",
   "eu",
   "other"
  ],
  "soc2": true,
  "hipaa": true,
  "sso": true,
  "lang_python": true,
  "lang_node": true,
  "lang_go": null,
  "lang_rust": null,
  "lang_java": null,
  "gpu_desktop": false,
  "credential_injection": null,
  "egress_http_rules": false,
  "wake_on_request": true,
  "live_resize": false,
  "fork_running": false,
  "memory_snapshot_fork": false,
  "webhooks": true,
  "scheduled_wakeups": null,
  "audit_logs": true,
  "eu_data_residency": true,
  "zero_data_retention": null,
  "scoped_api_keys": true,
  "spend_limits": true,
  "mcp_server": true,
  "snapshot_auto": false,
  "snapshot_on_demand": false,
  "harness_api": false,
  "ingress_rules": true,
  "secret_proxy": false,
  "volume_attach": true,
  "volume_shared": true,
  "isolation": "container",
  "image_oci": true,
  "anti_bot": false,
  "captcha_solving": false,
  "egress_ip_type": "datacenter",
  "own_agent_api": false,
  "guest_firewall": false,
  "max_session_unlimited": true,
  "open_internet": true,
  "usage_api": false,
  "persist_scope": "folder",
  "persist_path": "/workspace"
 },
 "perf": {
  "cpu_runs_s": null,
  "cold_start_ms": null,
  "burst_tti_ms": null,
  "dax_s": null
 },
 "perf_source": "No benchmark row: provider is not in ComputeSDK results or the HPC run (as of 2026-09-28). See research/perf-map.json.",
 "knobs": [
  {
   "id": "serverless_idle_timeout",
   "label": "Serverless billed time vs execution time (cold start + idle timeout)",
   "kind": "slider",
   "min": 1,
   "max": 2,
   "step": 0.05,
   "default": 1.1,
   "unit": "x",
   "options": null,
   "effect": "hours_factor",
   "modes": [
    "serverless-flex",
    "serverless-cpu"
   ],
   "why": "Flex workers bill from container start to full stop: cold start + execution + idle timeout (default 5 s). Short requests with heavy model loads can bill far more than execution time."
  },
  {
   "id": "retained_state_storage",
   "label": "Where retained state is stored",
   "kind": "select",
   "min": null,
   "max": null,
   "step": null,
   "default": 0.2,
   "unit": null,
   "options": [
    {
     "value": 0.2,
     "label": "Volume disk on stopped pod ($0.20/GB-mo)"
    },
    {
     "value": 0.07,
     "label": "Network volume < 1 TB ($0.07/GB-mo)"
    },
    {
     "value": 0.05,
     "label": "Network volume > 1 TB ($0.05/GB-mo)"
    },
    {
     "value": 0.14,
     "label": "High-performance network volume ($0.14/GB-mo)"
    }
   ],
   "effect": "snapshot_gib_month",
   "modes": null,
   "why": "Stopped-pod volume disk costs double the running rate; network volumes are cheaper and survive $0-balance termination. Rates from runpod.io/pricing and docs.runpod.io/pods/pricing (verified 2026-09-28)."
  }
 ],
 "caveats": [
  "CPU prices come from Runpod's public GraphQL catalogue (api.runpod.io/graphql cpuFlavors.specifics, callable without an API key; 2026-09-29), not from the pricing page (which lists GPUs only): cpu3c pod $0.03/vCPU-h, serverless $0.036/vCPU-h. The docs API v2 example ($0.04/$0.03) is stale/illustrative. GraphQL is scheduled for retirement in early 2027; REST v2 catalog needs a key.",
  "cpu3c RAM is 2 GB per vCPU (GraphQL cpuFlavors ramMultiplier 2; Flash IDs cpu3c-4-8); the 2.5 GB in the API v2 docs example is not the live value.",
  "GPU pods: per-GPU price bundles a fixed vCPU/RAM shape per GPU type; vcpu_h/ram_gib_h encoded as 0.",
  "Spot/interruptible: the Cloud GPUs product page FAQ still says 'We offer spot instances ... The UI/API will indicate current spot availability and pricing', but the docs no longer describe spot and the public GraphQL catalogue returns spot/bid prices equal to on-demand for every GPU (e.g. H100 SXM securePrice 3.49 = secureSpotPrice 3.49; lowestPrice minimumBidPrice = uninterruptablePrice). No spot discount is currently obtainable in the public data; not modelled.",
  "Savings plans: 3- or 6-month upfront prepay for GPU compute only (storage at standard rates), non-refundable, auto-applies to next deployment of the same GPU type; discount % not published (console only). Not modelled.",
  "Serverless active (always-on) workers: 'discounts available through sales inquiry', rate unpublished. Reserved clusters sales-only.",
  "Billing granularity: pods/pricing and billing overview say per second; the docs overview page says pods are 'billed by the minute' (conflict). Charges deducted every 5 min.",
  "Storage: container disk $0.10/GB-mo (running only, erased on stop); volume disk $0.10 running / $0.20 stopped; network volume $0.07 (<1 TB) / $0.05 (>1 TB), high-performance $0.14; network volumes billed hourly. storage.snapshot_gib_month = stopped volume-disk rate; knob selects network-volume alternative. Not charged while host is unavailable.",
  "No ingress/egress fees (official). Public IP/TCP ports available on many pods; no IPv4 price published (ipv4_month null).",
  "features.snapshot = 'fs' (verifier 2026-09-28, was 'none'): there is no snapshot API, but a stopped pod keeps its volume disk and network volumes persist, i.e. filesystem state is retained; with 'none' the engine refused every retained-state workload even though the card prices it.",
  "No idle auto-stop for pods: a running pod bills until you stop it. Stopped pods may restart with zero GPUs if the host's GPUs are taken.",
  "At $0 balance pods stop; pods without a network volume are terminated and data is lost.",
  "Knob retained_state_storage (effect 'snapshot_gib_month') overrides storage.snapshot_gib_month with the stopped-volume or network-volume rate.",
  "Compliance (runpod.io/legal/compliance, updated 2026-09-13): SOC 2 Type II and SOC 3 completed, ISO/IEC 27001:2022 certified (valid 2026-08-27 to 2029-08-26), HIPAA and GDPR programs; documents via Trust Center. SSO/SAML not found in docs (sso stays null). No benchmark data for Runpod in this research set.",
  "History: the RunPod \"slashes GPU prices\" blog post carries conflicting dates (HTML time 2024-07-12 vs visible \"Updated September 13, 2026\"); its old/new tables are not treated as a confirmed September 2026 cut. Current card rates come from the live pricing page. (patch v8-history, 2026-09-28)",
  "Instant Clusters (fix 2026-09-28): alt flag removed and gpu_min_count 16 set (2 nodes x 8 GPUs minimum per docs), so a 1-GPU request sees it only as a \"16-GPU nodes only\" compromise priced at 16 GPUs.",
  "Community Cloud is peer-hosted with variable reliability (no SLA stated); no engine flag exists for peer/marketplace capacity, so it is priced like Secure Cloud. Engine request filed (research/verify/engine-requests.md).",
  "CPU pods are plain containers you provision (no sandbox API); provisioning time is unpublished (boot_overhead_s null). Not flagged alt: a CPU pod is an ordinary on-demand machine; the engine already marks gpu-cloud rows \"plain VM / platform, no sandbox API\" for sandbox-API personas.",
  "Global Volumes (beta since 2026-09-15): region-independent object-backed storage, $0.09/GB-month plus request fees (Class A writes/lists $0.005 per 1,000, Class B reads $0.0005 per 1,000); deleted 15 days after the balance hits $0. More expensive than a standard network volume ($0.07) for parked state; not modelled.",
  "High-performance network volume: docs say \"Exact pricing varies by data center\"; $0.14/GB-month is the pricing-page figure (knob option), not a guaranteed rate in every DC.",
  "Startup Program: Starter tier $1,000 credits (application, not automatic); Growth tier = $50,000 upfront commit + $25,000 bonus credits under a 12-month agreement (effectively a 33% discount for $50k prepay). Referral sign-ups (Google SSO, after loading $10) get a $5-$500 random bonus (EU: $5; ~96% get <= $10). None modelled as free credit.",
  "Pricing page disagreement (2026-10-14): the visible Pods grid (\"GPU Cloud Pricing Updated October 8, 2026\", Secure Cloud toggle checked by default) shows the new Secure rates (H100 SXM $3.99, B200 $7.99, RTX 4090 $0.89), while the page's embedded schema.org Product data still says the old ones (\"H100 SXM GPU on Runpod ... Community Cloud $2.69/hr, Secure Cloud $3.49/hr\"). The card follows the dated, visible grid. Community rates are the same in both (grid data-community attributes = schema.org Community offers).",
  "Regions: the pricing page says Pods are \"deployed across 31 global regions\" (\"Thousands of GPUs across 30+ regions\"), but no public page lists 31 datacenters. locations now merges the Flash datacenter list (15) with the Global networking table (17, docs.runpod.io/pods/networking) = 24 datacenters incl. Canada (CA-MTL-3) and Australia (OC-AU-1). AP-JP-1 (Japan, announced in the March 2025 changelog) appears on no current datacenter list, so no Asia location is claimed; mode regions still carry \"asia\" from earlier research (unverified)."
 ],
 "sources": [
  "https://www.runpod.io/pricing",
  "https://docs.runpod.io/pods/pricing",
  "https://docs.runpod.io/serverless/pricing",
  "https://docs.runpod.io/accounts-billing/billing",
  "https://docs.runpod.io/accounts-billing/post-paid-billing",
  "https://docs.runpod.io/api-reference-v2/catalog/list-cpu-types",
  "https://docs.runpod.io/flash/configuration/cpu-types",
  "https://docs.runpod.io/flash/pricing",
  "https://docs.runpod.io/pods/choose-a-pod",
  "https://docs.runpod.io/llms-full.txt",
  "https://docs.runpod.io/instant-clusters",
  "https://docs.runpod.io/pods/networking"
 ],
 "fleet": {
  "default_max_instances": null,
  "default_max_vcpu": null,
  "default_max_gpus": null,
  "raise": "automatic",
  "raise_time_days": null,
  "verification": "Prepaid balance; getting started can use $10.",
  "tiers": "Default aggregate spend cap $80/hour, growing with account history (or via support). Serverless max workers (flex + active combined) by balance: default 5, $100+ 10, $200+ 20, $300+ 30, $500+ 40, $700+ 50, $900+ 60; more via support. Instant Clusters: 2 nodes/16 GPUs self-serve, up to 8 nodes/64 GPUs after a spend-limit increase.",
  "notes": "Pods card: no numeric default Pod/GPU count established. Do not turn serverless worker limits or the dollar cap into a fixed machine count; hardware rate and stock determine attainable fleet.",
  "sources": [
   "https://github.com/runpod/docs/blob/main/accounts-billing/billing.mdx",
   "https://contact.runpod.io/hc/en-us/articles/50501627768595-Increase-Serverless-Worker-Limits"
  ],
  "researched": "2026-09-28"
 },
 "audit_2026_09_29": {
  "applied": 13,
  "source": "checked against the provider's live pricing and docs pages",
  "reviewed": 14
 },
 "feature_evidence": {
  "snapshot_auto": {
   "value": false,
   "source": "https://docs.runpod.io/pods/manage-pods.md",
   "quote": "Stopping a Pod releases the GPU and preserves data in `/workspace` (volume disk). Container disk data is cleared.",
   "note": "Stop keeps only the volume disk / network volume (persistent_disk), and the container disk is wiped; no state capture, no memory. Pods have no idle auto-stop.",
   "checked": "2026-09-29"
  },
  "snapshot_on_demand": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "no snapshot/checkpoint/save-image feature anywhere in the docs; the API v2 Pod resource has only list/create/get/update/terminate/logs/state-transition (start/stop/restart)",
   "note": "No on-demand snapshot/checkpoint API exists for Pods or Serverless. The apparent contradiction is that the card's `snapshot` field is set to \"fs\" only because a stopped Pod keeps its volume/network disk (persistent_disk-style retention), not because a real filesystem-snapshot API exists (that field, not snapshot_on_demand, is the mismatch to flag). snapshot_on_demand=false stands.",
   "checked": "2026-09-30"
  },
  "harness_api": {
   "value": false,
   "source": "https://docs.runpod.io/get-started/agent-skills.md",
   "quote": "Official Runpod instructions that install the Runpod skills and MCP server into your coding agent in one shot.",
   "note": "Runpod's agent features (skills plugin, MCP server at mcp.getrunpod.io) let YOUR local Claude Code/Codex manage Runpod resources; nothing drives an agent harness inside a Pod. Serverless job API is a generic handler queue, not a harness API.",
   "checked": "2026-09-29"
  },
  "ingress_rules": {
   "value": true,
   "source": "https://docs.runpod.io/pods/configuration/expose-ports.md",
   "quote": "To configure HTTP ports during Pod deployment, click **Edit Template** and add a comma-separated list of ports to the **Expose HTTP Ports (Max 10)** field.",
   "note": "Control is limited to choosing which ports are exposed (HTTP proxy https://POD_ID-PORT.proxy.runpod.net, max 10, or direct TCP on a public IP) plus a private-only option: Global networking gives each Pod a private IP 'accessible only to other Pods in your account' (NVIDIA GPU Pods, 17 DCs). Exposed ports are fully public: 'Your service becomes publicly accessible. While the Pod ID provides some obscurity, implement proper authentication in your application.' No auth on proxy URLs, no IP allowlists/security groups.",
   "checked": "2026-09-29"
  },
  "secret_proxy": {
   "value": false,
   "source": "https://docs.runpod.io/pods/templates/secrets.md",
   "quote": "Reference your secrets directly in the [environment variables section](/pods/templates/environment-variables) of your Pod template using the `RUNPOD_SECRET_` prefix followed by your secret name",
   "note": "Runpod secrets are encrypted at rest and write-only in the console/API, but are delivered into the Pod as plain environment variables. No injecting proxy.",
   "checked": "2026-09-29"
  },
  "volume_attach": {
   "value": true,
   "source": "https://docs.runpod.io/pods/storage/types.md",
   "quote": "Network volumes provide permanent storage that exists independently from any Pod. You can attach a network volume to multiple Pods, transfer it between machines, and retain your data even after deleting a Pod.",
   "note": "Per-Pod volume disk (/workspace, survives stop, deleted on terminate) plus network volumes (Secure Cloud only; 'must be attached during Pod creation and cannot be detached later'; replaces the volume disk at /workspace) and global volumes (beta, region-independent, read-optimized). One network volume + one global volume per Pod. Also attachable to Serverless endpoints (/runpod-volume) and Instant Clusters.",
   "checked": "2026-09-29"
  },
  "volume_shared": {
   "value": true,
   "source": "https://docs.runpod.io/storage/network-volumes.md",
   "quote": "Writing to the same volume from multiple workers simultaneously may cause data corruption. Handle concurrent write access in your application logic.",
   "note": "Read-write, mounted by multiple running Pods/workers at once within one datacenter (storage table: 'Shareable ... Yes (across Pods)'); no locking/consistency guarantees. Global volumes (beta) are shareable 'across Pods in any data center'. S3-compatible API to network volumes too.",
   "checked": "2026-09-29"
  },
  "isolation": {
   "value": "container",
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "Your Pods and workers run in a multi-tenant environment with containerized isolation that prevents other users from accessing your data. Each Pod/worker operates in its own container with strict separation from other workloads.",
   "note": "Docs explicitly describe container-based multi-tenant isolation (shared kernel), matching the card's top-level isolation:container; consistent with docker_inside:false (no nested Docker/Compose inside a Pod).",
   "checked": "2026-09-30"
  },
  "image_oci": {
   "value": true,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "Custom Docker image to deploy. When specified, the endpoint runs your Docker image instead of Flash's managed workers.",
   "note": "Pods and Serverless endpoints both start from custom Docker/OCI images (public or private registries); templates package a Docker image + config for reuse.",
   "checked": "2026-09-30"
  },
  "nested_virt": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "Each container includes its own dependencies... Containers share the host OS kernel, using fewer resources than virtual machines.",
   "note": "Pods/workers are containers sharing the host kernel; no /dev/kvm, nested-virtualization, or VM-inside-Pod capability is documented anywhere in the docs corpus searched.",
   "checked": "2026-09-30"
  },
  "anti_bot": {
   "value": false,
   "source": "https://www.runpod.io/pricing",
   "quote": "Runpod is a cloud platform for building AI applications... Pods, Serverless GPU endpoints, and AI infrastructure.",
   "note": "not offered: outside the product's scope (GPU/CPU cloud compute platform, not a browser/scraping product; no anti-bot/stealth or fingerprinting feature found in docs).",
   "checked": "2026-09-30"
  },
  "captcha_solving": {
   "value": false,
   "source": "https://www.runpod.io/pricing",
   "quote": "Runpod is a cloud platform for building AI applications... Pods, Serverless GPU endpoints, and AI infrastructure.",
   "note": "not offered: outside the product's scope (GPU/CPU cloud compute platform; no CAPTCHA-solving feature found in docs).",
   "checked": "2026-09-30"
  },
  "egress_ip_type": {
   "value": "datacenter",
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "Public IP addresses may change for Community Cloud Pods if your Pod is migrated or restarted, but they should remain stable for Secure Cloud Pods.",
   "note": "Standard datacenter IPs assigned per Pod/host; no residential or mobile IP product is offered.",
   "checked": "2026-09-30"
  },
  "wake_on_request": {
   "value": true,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "If no workers are active, Runpod automatically starts one (cold start). The request is queued until a worker is available.",
   "note": "Applies to Serverless workers scaled to zero (workers=(0,n)); Pods themselves have no idle auto-stop/wake (must be started manually), so this reflects the Serverless product only.",
   "checked": "2026-09-30"
  },
  "live_resize": {
   "value": false,
   "source": "https://docs.runpod.io/pods/manage-pods.md",
   "quote": "Stopping a Pod releases the GPU and preserves data in `/workspace` (volume disk). Container disk data is cleared.",
   "note": "No documented API/console feature to change a running Pod's CPU/RAM/GPU without stopping and reconfiguring it; changing size requires stop/edit/restart, which releases the GPU.",
   "checked": "2026-09-30"
  },
  "own_agent_api": {
   "value": false,
   "source": "https://docs.runpod.io/get-started/agent-skills.md",
   "quote": "Official Runpod instructions that install the Runpod skills and MCP server into your coding agent in one shot.",
   "note": "Runpod's agent-facing features (skills plugin, MCP server at mcp.getrunpod.io) let a user's own coding agent/LLM manage Runpod infrastructure; Runpod does not ship its own browser/computer-use/research task agent with a driving API.",
   "checked": "2026-09-30"
  },
  "lang_node": {
   "value": true,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "The JavaScript SDK supports Node.js and modern JavaScript environments using ES modules.",
   "note": "First-party runpod-sdk npm package for Node.js/JS, alongside the Python SDK.",
   "checked": "2026-09-30"
  },
  "egress_allowlist": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "Ensure your network allows outbound connections to the Runpod IP range",
   "note": "No provider-side outbound allow/deny list, no-internet mode, or security-group egress rules found anywhere in the docs corpus; the only related text is about the user's own network permitting outbound connections to Runpod, not the reverse.",
   "checked": "2026-09-30"
  },
  "egress_http_rules": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "A firewall or VPN configured with a specific Pod ID. A firewall or VPN configured with a specific Pod IP address.",
   "note": "The only firewall/rule mentions are examples of the customer's own external firewall/VPN pointed at a Pod; no provider-side HTTP method/path-level egress control feature is documented.",
   "checked": "2026-09-30"
  },
  "guest_firewall": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "Containers share the host OS kernel, using fewer resources than virtual machines.",
   "note": "derived: Pods/workers are containers (shared kernel), not a full VM/microVM, and no capability grants (NET_ADMIN) for iptables/nftables are documented, so guest-side firewalling is not assumed to work despite root access.",
   "checked": "2026-09-30"
  },
  "static_egress_ip": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "Public IP addresses may change for Community Cloud Pods if your Pod is migrated or restarted, but they should remain stable for Secure Cloud Pods.",
   "note": "IPs are only 'stable' while a Secure Cloud Pod keeps running, not a reserved/fixed IP product; Community Cloud IPs can change on migration/restart, and no dedicated static-IP purchase option is documented.",
   "checked": "2026-09-30"
  },
  "sso": {
   "value": true,
   "source": "https://www.runpod.io/enterprise",
   "quote": "SSO/SAML Single sign-on with your existing identity provider, Okta, Entra ID, Auth0, Google Workspace, or Duo. Our team sets it up with you directly.",
   "note": "Available on the Organizations/Enterprise tier, set up directly with Runpod's team; not self-service in the standard console.",
   "checked": "2026-09-30"
  },
  "max_session_h": {
   "value": "none",
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "Pods are billed by the second for compute and storage, with no fees for data ingress or egress.",
   "note": "Pods have no documented maximum runtime or forced stop; a running Pod bills and keeps running until the user stops it or account balance hits $0 (auto_stop_idle:false in the card). Serverless workers are governed by per-job TTL/executionTimeout (default 24h, up to 7 days), not a session cap on the product itself.",
   "checked": "2026-09-30"
  },
  "fork": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "click the three dots to the top right of your production endpoint. Click Clone Endpoint. ... Each environment maintains independent GPU and worker configurations.",
   "note": "Only \"clone\" concept in the docs is Clone Endpoint for Serverless (copies config/repository settings to a new, empty endpoint) and 'Save as template' (image+config only). No API/console action clones a running Pod's disk/state into a new machine.",
   "checked": "2026-09-30"
  },
  "fork_running": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "Click the three dots to the top right of your production endpoint. Click Clone Endpoint. Expand the Repository Configuration section and select your staging branch.",
   "note": "Same evidence as fork: cloning duplicates configuration, not a live running machine's memory/disk state. No fork-while-running feature exists.",
   "checked": "2026-09-30"
  },
  "memory_snapshot_fork": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "no snapshot/checkpoint/save-image feature anywhere in the docs",
   "note": "There is no memory-state capture of any kind (see snapshot_on_demand), so a clone that preserves running processes/memory is impossible; consistent with fork/fork_running being false.",
   "checked": "2026-09-30"
  },
  "docker_inside": {
   "value": false,
   "source": "https://docs.runpod.io/pods/overview.md",
   "quote": "Docker Compose is not supported: Runpod runs Docker for you, so you cannot spin up your own Docker instance or use Docker Compose on Pods.",
   "note": "Pods are themselves containers; nested Docker (docker-in-docker) and Docker Compose are explicitly unsupported.",
   "checked": "2026-09-30"
  },
  "desktop": {
   "value": false,
   "source": "https://docs.runpod.io/pods/connect-to-a-pod.md",
   "quote": "You can connect directly to your running Pods and execute code on them using a variety of methods, including a built-in web terminal, an SSH connection from your local machine, a JupyterLab instance, or a remote VSCode/Cursor development environment.",
   "note": "Only text-based/terminal, SSH, Jupyter and remote-IDE access are offered; no VNC/RDP/graphical-desktop product anywhere in the docs.",
   "checked": "2026-09-30"
  },
  "browser": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "Web Terminal: A browser-based terminal.",
   "note": "\"Browser\" mentions in the docs are the browser-based terminal/Jupyter UI and the CLI's OAuth login flow, not a managed/real browser instance offered as a product feature.",
   "checked": "2026-09-30"
  },
  "computer_use": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "no snapshot/checkpoint/save-image feature anywhere in the docs; the API v2 Pod resource has only list/create/get/update/terminate/logs/state-transition (start/stop/restart)",
   "note": "No desktop exists to drive (see `desktop`), and no screenshot+click / computer-use API appears anywhere in the docs or API reference. Runpod's product is GPU/CPU compute pods and serverless workers, not a desktop-automation service.",
   "checked": "2026-09-30"
  },
  "auto_stop_idle": {
   "value": false,
   "source": "https://docs.runpod.io/pods/manage-pods.md",
   "quote": "Schedule a stop (e.g., after 2 hours): sleep 2h; runpodctl pod stop $RUNPOD_POD_ID &",
   "note": "Pods only stop when the user manually stops them (console/CLI/API) or must self-schedule a timer inside the Pod; there is no provider-side idle-detection auto-stop for Pods. (Serverless workers do scale to zero via idle_timeout, but that is captured per-mode as wake_on_request, not this top-level Pod feature.)",
   "checked": "2026-09-30"
  },
  "agent_harnesses": {
   "value": false,
   "source": "https://docs.runpod.io/get-started/agent-skills.md",
   "quote": "Official Runpod instructions that install the Runpod skills and MCP server into your coding agent in one shot.",
   "note": "This installs Runpod's MCP server/skills into the USER's own local coding agent (Claude Code, Codex, etc.) so it can manage Runpod resources; it is not a preinstalled coding-agent template running inside a Pod. No template list (PyTorch, Stable Diffusion, ComfyUI, Jupyter, etc.) includes a coding-agent harness.",
   "checked": "2026-09-30"
  },
  "windows": {
   "value": false,
   "source": "https://docs.runpod.io/pods/overview.md",
   "quote": "Windows is not supported: Pods do not currently support Windows.",
   "note": "Explicit limitation in the Pods overview page.",
   "checked": "2026-09-30"
  },
  "macos": {
   "value": false,
   "source": "https://docs.runpod.io/pods/overview.md",
   "quote": "Secure Cloud: Operates in T3/T4 data centers, providing high reliability and security for enterprise and production workloads.",
   "note": "Runpod is a Linux-container GPU/CPU cloud (Pods run Docker images in Runpod-operated data centers); macOS machines are outside the product's scope, consistent with `windows` also being unsupported.",
   "checked": "2026-09-30"
  },
  "arm64": {
   "value": false,
   "source": "https://docs.runpod.io/tutorials/introduction/containers/create-dockerfiles.md",
   "quote": "Runpod's infrastructure uses linux/amd64 architecture. If you're building on an Apple Silicon Mac (ARM64), specify the platform: docker build --platform=linux/amd64 -t my-image .",
   "note": "Runpod explicitly tells ARM (Apple Silicon) users to cross-build for amd64 because the infrastructure itself is amd64 only; no ARM64 compute is offered.",
   "checked": "2026-09-30"
  },
  "self_host": {
   "value": false,
   "source": "https://docs.runpod.io/pods/overview.md",
   "quote": "Runpod offers two cloud options: Secure Cloud: Operates in T3/T4 data centers... Community Cloud: Connects individual compute providers to users through a vetted, secure peer-to-peer system",
   "note": "Runpod is a fully managed multi-tenant cloud (its own or partner data centers); no docs describe running the Runpod platform itself on a customer's own infrastructure.",
   "checked": "2026-09-30"
  },
  "byoc": {
   "value": false,
   "source": "https://docs.runpod.io/pods/overview.md",
   "quote": "Runpod offers two cloud options: Secure Cloud... Community Cloud...",
   "note": "Both offered cloud options are Runpod-operated (own data centers or peer-hosted capacity onboarded by Runpod); no bring-your-own-cloud-account deployment mode is documented anywhere in llms-full.txt.",
   "checked": "2026-09-30"
  },
  "open_source": {
   "value": false,
   "source": "https://docs.runpod.io/llms-full.txt",
   "quote": "runpodctl - Pre-built binaries are available on conda-forge for Linux (x86_64, aarch64, ppc64le), macOS (x86_64, arm64), and Windows (x86_64).",
   "note": "Only peripheral tooling (the runpodctl CLI, worker templates on GitHub) is open source; the core Pods/Serverless platform, console, and billing engine are Runpod's proprietary closed product with no OSI license mentioned anywhere in the docs.",
   "checked": "2026-09-30"
  },
  "usage_api": {
   "value": false,
   "source": "https://docs.runpod.io/accounts-billing/billing",
   "quote": "You can view your current balance and recent transactions on the Billing page at any time. [...] Review recent spending in the Billing Explorer section. For detailed cost attribution by team or project, see Cost centers.",
   "note": "granularity: account-only; cost: yes; RunPod offers console-based billing tracking through the Billing page and cost centers (which attach labels to resources for cost attribution), but no REST API or CLI endpoint to programmatically retrieve billable usage (compute seconds, vCPU-h, GB-h, or cost) for individual sandboxes/pods. The API v2 provides pod metadata and management but not usage/billing data. Cost centers enable billing labels for organizational tracking, but remain a console feature without programmatic access.",
   "confidence": "high",
   "checked": "2026-10-01"
  },
  "persist_scope": {
   "value": "folder",
   "source": "https://docs.runpod.io/pods/manage-pods.md",
   "quote": "Stopping a Pod releases the GPU and preserves data in `/workspace` (volume disk). Container disk data is cleared.",
   "note": "Only /workspace survives a stop; the rest of the filesystem (packages installed elsewhere) resets.",
   "checked": "2026-10-01"
  }
 },
 "locations": [
  {
   "country": "US",
   "state": "CA",
   "city": "Los Angeles",
   "provider_region": "us-ca-2",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "GA",
   "city": "Atlanta",
   "provider_region": "us-ga-2",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "IL",
   "city": "Chicago",
   "provider_region": "us-il-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "KS",
   "city": "Kansas City",
   "provider_region": "us-ks-2",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "MD",
   "city": "Baltimore",
   "provider_region": "us-md-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "MO",
   "city": "St. Louis",
   "provider_region": "us-mo-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "MO",
   "city": "Kansas City",
   "provider_region": "us-mo-2",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "NC",
   "city": "Charlotte",
   "provider_region": "us-nc-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "NC",
   "city": "Raleigh",
   "provider_region": "us-nc-2",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "NE",
   "city": "Omaha",
   "provider_region": "us-ne-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "WA",
   "city": "Seattle",
   "provider_region": "us-wa-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "CZ",
   "state": null,
   "city": "Prague",
   "provider_region": "eu-cz-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "RO",
   "state": null,
   "city": "Bucharest",
   "provider_region": "eu-ro-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "IS",
   "state": null,
   "city": "Reykjavik",
   "provider_region": "eur-is-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "NO",
   "state": null,
   "city": "Oslo",
   "provider_region": "eur-no-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "CA",
   "state": null,
   "city": null,
   "provider_region": "ca-mtl-3",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "FR",
   "state": null,
   "city": null,
   "provider_region": "eu-fr-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "NL",
   "state": null,
   "city": null,
   "provider_region": "eu-nl-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "SE",
   "state": null,
   "city": null,
   "provider_region": "eu-se-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "IS",
   "state": null,
   "city": null,
   "provider_region": "eur-is-2",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "AU",
   "state": null,
   "city": null,
   "provider_region": "oc-au-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "GA",
   "city": null,
   "provider_region": "us-ga-1",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "TX",
   "city": null,
   "provider_region": "us-tx-3",
   "self_serve": true,
   "choosable": true
  },
  {
   "country": "US",
   "state": "TX",
   "city": null,
   "provider_region": "us-tx-4",
   "self_serve": true,
   "choosable": true
  }
 ],
 "locations_source": "https://docs.runpod.io/llms-full.txt",
 "locations_quote": "Available datacenters: US regions (CA, GA, IL, KS, MD, MO x2, NC x2, NE, WA) and EU regions (CZ, RO, IS, NO). Datacenters can be selected during Pod/Serverless deployment via the datacenter parameter.",
 "locations_checked": "2026-10-14",
 "locations_note": "First 15 from the Flash datacenter list (llms-full.txt); the other 9 (CA-MTL-3, EU-FR-1, EU-NL-1, EU-SE-1, EUR-IS-2, OC-AU-1, US-GA-1, US-TX-3, US-TX-4) from the Global networking table, which names only the country/state: \"Global networking is available in these 17 data centers worldwide\" (https://docs.runpod.io/pods/networking). The pricing page claims 31 regions; the rest are unlisted."
}