{
 "features_resolved": {
  "snapshot_any": false,
  "snapshot_mem": false,
  "fork": false,
  "pause_resume": false,
  "persistent_disk": true,
  "whole_disk": false,
  "persist_scope": true,
  "volumes": true,
  "auto_stop_idle": true,
  "long_sessions": true,
  "custom_image": true,
  "image_snapshot": false,
  "image_oci": true,
  "vm_isolation": false,
  "docker_inside": true,
  "nested_virt": false,
  "root": null,
  "systemd": null,
  "browser": false,
  "desktop": false,
  "computer_use": false,
  "code_interpreter": null,
  "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": null,
  "egress_allowlist": false,
  "open_internet": null,
  "static_egress_ip": false,
  "private_network": null,
  "self_host": false,
  "byoc": false,
  "open_source": false,
  "soc2": true,
  "hipaa": false,
  "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": null,
  "credential_injection": null,
  "egress_http_rules": false,
  "wake_on_request": true,
  "live_resize": false,
  "fork_running": false,
  "memory_snapshot_fork": false,
  "webhooks": null,
  "scheduled_wakeups": null,
  "audit_logs": null,
  "eu_data_residency": false,
  "zero_data_retention": null,
  "scoped_api_keys": null,
  "spend_limits": null,
  "mcp_server": true,
  "usage_api": false,
  "snapshot_auto": false,
  "snapshot_on_demand": false,
  "harness_api": false,
  "own_agent_api": true,
  "fast_boot": null,
  "strong_isolation": false,
  "ingress_rules": true,
  "guest_firewall": true,
  "secret_proxy": false,
  "secret_proxy_any": false,
  "volume_attach": true,
  "volume_shared": true,
  "sidecar": null,
  "fuse": null,
  "kernel_features": null,
  "burst": null
 },
 "company": {
  "id": "fal",
  "checked": "2026-09-30",
  "company": "fal.ai (Features and Labels, Inc.)",
  "ownership": {
   "value": "independent",
   "parent": null,
   "source": "https://techcrunch.com/2025/10/21/sources-multimodal-ai-startup-fal-ai-already-raised-at-4b-valuation/",
   "quote": "The startup was co-founded in 2021 by Burkay Gur, a former Coinbase machine learning leader and Oracle engineer, and Gorkem Yurtseven, who was previously a developer at Amazon."
  },
  "stage": {
   "value": "series-c",
   "source": "https://techcrunch.com/2025/10/21/sources-multimodal-ai-startup-fal-ai-already-raised-at-4b-valuation/",
   "quote": "The new round is coming less than three months after Fal announced a $125 million Series C at a $1.5 billion valuation led by Meritech.",
   "confidence": "medium",
   "note": "TechCrunch (Oct 2025) reports fal 'closed a new round' of ~$250M at a >$4B valuation led by Kleiner Perkins and Sequoia, but does not name the round's series letter (likely Series D); labeled here as series-c because that is the last round TechCrunch explicitly named as a stage. If the >$4B round is confirmed as a priced Series D, stage would be series-d+."
  },
  "accelerators": [],
  "total_funding_usd": {
   "value": 448000000,
   "status": "estimated",
   "basis": "Sum of confirmed TechCrunch-reported rounds: $9M seed + $14M Series A (2024) + $125M Series C (Jul 2025) + ~$250M unnamed round (Oct 2025, TechCrunch citing 4 sources); PitchBook separately estimated ~$200M raised prior to the Oct 2025 round per TechCrunch.",
   "source": "https://techcrunch.com/2025/10/21/sources-multimodal-ai-startup-fal-ai-already-raised-at-4b-valuation/",
   "quote": "Fal previously raised nearly $200 million, according to PitchBook data."
  },
  "last_round": {
   "type": "series-d+ (unconfirmed label)",
   "amount_usd": 250000000,
   "date": "2025-10",
   "leads": [
    "Kleiner Perkins",
    "Sequoia"
   ],
   "status": "confirmed",
   "source": "https://techcrunch.com/2025/10/21/sources-multimodal-ai-startup-fal-ai-already-raised-at-4b-valuation/",
   "quote": "Fal.ai, a startup that hosts image, video, and audio AI models for developers, has closed a new round valuing the company at over $4 billion, four people familiar with the deal said. The company raised approximately $250 million, two of the people said."
  },
  "investors": [
   "Kleiner Perkins",
   "Sequoia",
   "Meritech Capital",
   "Bessemer Venture Partners",
   "Andreessen Horowitz (a16z)",
   "Kindred Ventures",
   "Notable Capital",
   "First Round Capital",
   "Unusual Ventures",
   "Village Global"
  ],
  "portfolio_listings": [
   {
    "source": "First Round Capital",
    "company": "Fal",
    "batch": null,
    "how": "domain"
   },
   {
    "source": "Unusual",
    "company": "Fal",
    "batch": null,
    "how": "domain"
   },
   {
    "source": "Village Global",
    "company": "Fal.AI",
    "batch": null,
    "how": "domain"
   }
  ],
  "revenue": {
   "value_usd": 95000000,
   "kind": "ARR",
   "as_of": "2025-07",
   "status": "confirmed",
   "basis": "TechCrunch citing First Round Capital partner Todd Jackson's LinkedIn post at the time of the Series C announcement",
   "source": "https://techcrunch.com/2025/10/21/sources-multimodal-ai-startup-fal-ai-already-raised-at-4b-valuation/",
   "quote": "At that time, the company's revenue crossed $95 million and its platform was used by over 2 million developers, Todd Jackson, a partner at First Round Capital, wrote on LinkedIn."
  },
  "employees": {
   "value": null,
   "source": "https://techcrunch.com/2024/09/18/fal-ai-which-hosts-media-generating-ai-models-raises-23m-from-a16z-and-others/"
  },
  "note": "Sep 2024 TechCrunch article reported Fal's annual run rate at nearly $10M (~$800k/month) and a 17-person staff, with the Series A valuing the company at $80M; by Jul 2025 ARR reportedly reached ~$95M at a $1.5B valuation (Series C, led by Meritech), and by Oct 2025 the company reportedly raised ~$250M more at a >$4B valuation led by Kleiner Perkins/Sequoia (fal itself did not comment). Not a YC company (ycombinator.com/companies/fal returns 404); no accelerator/incubator affiliation found. Co-founded 2021 by Burkay Gur and Gorkem Yurtseven."
 },
 "id": "fal",
 "name": "fal",
 "url": "https://fal.ai/pricing",
 "category": "gpu-cloud",
 "isolation": "container",
 "modes": [
  {
   "key": "serverless-b300-0",
   "label": "B300, serverless (×1)",
   "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": null,
   "granularity_s": 1,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": {
    "B300": 8.5
   },
   "note": " Serverless machine bundle. CPU listed as vendor CPU count, physical-vs-logical not stated. B200 docs say 192GB VRAM vs pricing 180GB; host RAM is separate. Source: https://fal.ai/pricing",
   "product_class": "gpu-instance"
  },
  {
   "key": "reserved-b300-1",
   "label": "B300, reserved",
   "flags": [
    "sales"
   ],
   "os": [
    "linux"
   ],
   "regions": null,
   "gpu_only": true,
   "features": {},
   "pricing": "resource",
   "vcpu_h": null,
   "ram_gib_h": null,
   "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": true,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": null,
   "granularity_s": null,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": null,
   "note": "As-low-as custom deployment quote; spend/term not published, requires support. Source: https://fal.ai/pricing ESTIMATOR EXCLUDED: Not a fixed verified USD quote. Bundled node count/CPU/RAM constraints are not enforceable in current engine schema. CPU/RAM inclusion not verified. Exact researched rate remains in gpu.json."
  },
  {
   "key": "serverless-b200-2",
   "label": "B200, serverless (×1)",
   "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": 19,
   "max_ram_gib": 210,
   "sizes": null,
   "pool_tiers": null,
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": null,
   "granularity_s": 1,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": {
    "B200": 6.25
   },
   "note": " Serverless machine bundle. CPU listed as vendor CPU count, physical-vs-logical not stated. B200 docs say 192GB VRAM vs pricing 180GB; host RAM is separate. Source: https://fal.ai/pricing",
   "product_class": "gpu-instance"
  },
  {
   "key": "reserved-b200-3",
   "label": "B200, reserved",
   "flags": [
    "sales"
   ],
   "os": [
    "linux"
   ],
   "regions": null,
   "gpu_only": true,
   "features": {},
   "pricing": "resource",
   "vcpu_h": null,
   "ram_gib_h": null,
   "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": true,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": null,
   "granularity_s": null,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": null,
   "note": "As-low-as custom deployment quote; spend/term not published, requires support. Source: https://fal.ai/pricing ESTIMATOR EXCLUDED: Not a fixed verified USD quote. Bundled node count/CPU/RAM constraints are not enforceable in current engine schema. CPU/RAM inclusion not verified. Exact researched rate remains in gpu.json."
  },
  {
   "key": "serverless-h200-4",
   "label": "H200, serverless (×1)",
   "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": 12,
   "max_ram_gib": 112,
   "sizes": null,
   "pool_tiers": null,
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": null,
   "granularity_s": 1,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": {
    "H200": 4.5
   },
   "note": " Serverless machine bundle. CPU listed as vendor CPU count, physical-vs-logical not stated. B200 docs say 192GB VRAM vs pricing 180GB; host RAM is separate. Source: https://fal.ai/pricing",
   "product_class": "gpu-instance"
  },
  {
   "key": "reserved-h200-5",
   "label": "H200, reserved",
   "flags": [
    "sales"
   ],
   "os": [
    "linux"
   ],
   "regions": null,
   "gpu_only": true,
   "features": {},
   "pricing": "resource",
   "vcpu_h": null,
   "ram_gib_h": null,
   "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": true,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": null,
   "granularity_s": null,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": null,
   "note": "As-low-as custom deployment quote; spend/term not published, requires support. Source: https://fal.ai/pricing ESTIMATOR EXCLUDED: Not a fixed verified USD quote. Bundled node count/CPU/RAM constraints are not enforceable in current engine schema. CPU/RAM inclusion not verified. Exact researched rate remains in gpu.json."
  },
  {
   "key": "serverless-h100-6",
   "label": "H100, serverless (×1)",
   "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": 12,
   "max_ram_gib": 112,
   "sizes": null,
   "pool_tiers": null,
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": null,
   "granularity_s": 1,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": {
    "H100": 4.5
   },
   "note": " Serverless machine bundle. CPU listed as vendor CPU count, physical-vs-logical not stated. B200 docs say 192GB VRAM vs pricing 180GB; host RAM is separate. Source: https://fal.ai/pricing",
   "product_class": "gpu-instance"
  },
  {
   "key": "reserved-h100-7",
   "label": "H100, reserved",
   "flags": [
    "sales"
   ],
   "os": [
    "linux"
   ],
   "regions": null,
   "gpu_only": true,
   "features": {},
   "pricing": "resource",
   "vcpu_h": null,
   "ram_gib_h": null,
   "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": true,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": null,
   "granularity_s": null,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": null,
   "note": "As-low-as custom deployment quote; spend/term not published, requires support. Source: https://fal.ai/pricing ESTIMATOR EXCLUDED: Not a fixed verified USD quote. Bundled node count/CPU/RAM constraints are not enforceable in current engine schema. CPU/RAM inclusion not verified. Exact researched rate remains in gpu.json."
  },
  {
   "key": "serverless-rtx-pro-6000-8",
   "label": "RTX-PRO-6000, serverless (×1)",
   "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": 6,
   "max_ram_gib": 100,
   "sizes": null,
   "pool_tiers": null,
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": null,
   "granularity_s": 1,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": {
    "RTX-PRO-6000": 2.99
   },
   "note": " Serverless machine bundle. CPU listed as vendor CPU count, physical-vs-logical not stated. B200 docs say 192GB VRAM vs pricing 180GB; host RAM is separate. Source: https://fal.ai/pricing",
   "product_class": "gpu-instance"
  },
  {
   "key": "reserved-rtx-pro-6000-9",
   "label": "RTX-PRO-6000, reserved",
   "flags": [
    "sales"
   ],
   "os": [
    "linux"
   ],
   "regions": null,
   "gpu_only": true,
   "features": {},
   "pricing": "resource",
   "vcpu_h": null,
   "ram_gib_h": null,
   "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": true,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": null,
   "granularity_s": null,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": null,
   "note": "As-low-as custom deployment quote; spend/term not published, requires support. Source: https://fal.ai/pricing ESTIMATOR EXCLUDED: Not a fixed verified USD quote. Bundled node count/CPU/RAM constraints are not enforceable in current engine schema. CPU/RAM inclusion not verified. Exact researched rate remains in gpu.json."
  },
  {
   "key": "on-demand-dedicated-compute-sxm-10",
   "label": "H100 SXM, on-demand (×1)",
   "flags": [
    "alt",
    "sales"
   ],
   "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": 16,
   "max_ram_gib": 200,
   "sizes": null,
   "pool_tiers": null,
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": null,
   "granularity_s": 3600,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": null,
   "note": "Dedicated continuous Compute VM, not Serverless. 1TB SSD/GPU. Price visible only in dashboard/contact team; do not substitute serverless list or as-low-as quote. Source: https://fal.ai/docs/documentation/compute/pricing.md"
  },
  {
   "key": "on-demand-dedicated-compute-sxm-11",
   "label": "H100 SXM, on-demand (×8 node)",
   "flags": [
    "alt",
    "sales"
   ],
   "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": 128,
   "max_ram_gib": 1600,
   "sizes": null,
   "pool_tiers": null,
   "requires_always_on": false,
   "always_on_month_per_instance": null,
   "commit_note": null,
   "min_billed_seconds": null,
   "granularity_s": 3600,
   "start_fee": null,
   "boot_overhead_s": null,
   "multiplier": 1,
   "windows_vcpu_h": null,
   "gpu": null,
   "note": "Dedicated continuous Compute VM, not Serverless. 1TB SSD/GPU. Price visible only in dashboard/contact team; do not substitute serverless list or as-low-as quote. Source: https://fal.ai/docs/documentation/compute/pricing.md ESTIMATOR EXCLUDED: Bundled node count/CPU/RAM constraints are not enforceable in current engine schema. Exact researched rate remains in gpu.json."
  },
  {
   "key": "serverless-gb200",
   "label": "GB200, serverless (x1)",
   "flags": [],
   "os": [
    "linux"
   ],
   "regions": null,
   "gpu_only": true,
   "features": {},
   "pricing": "resource",
   "vcpu_h": 0,
   "ram_gib_h": 0,
   "cpu_basis": "alloc",
   "ram_basis": "alloc",
   "min_billed_seconds": null,
   "granularity_s": 1,
   "start_fee": null,
   "multiplier": 1,
   "gpu": {
    "GB200": 9.99
   },
   "note": "Listed only on fal.ai/serverless (not on fal.ai/pricing or machine-types docs); 'As low as $5.89'. CPU/RAM bundle unpublished. Source: https://fal.ai/serverless"
  }
 ],
 "storage": {
  "included_disk_gib": null,
  "disk_gib_month": null,
  "disk_billed_when_stopped": null,
  "snapshot_gib_month": null,
  "paused_ram_billed": null
 },
 "network": {
  "egress_free_gib": null,
  "egress_gib": null,
  "ipv4_month": null,
  "ipv4_available": null
 },
 "plans": [
  {
   "name": "Pay as you go (serverless, account approval)",
   "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": "List GPU prices on fal.ai/pricing, billed per second while runners are alive (SETUP/IDLE/RUNNING/DRAINING). Free account; serverless deploy access is approved per account by fal (docs: \"Serverless access is approved per account\"). Concurrency quotas not published."
  },
  {
   "name": "Enterprise / custom deployment",
   "fee": null,
   "flags": [
    "sales"
   ],
   "fee_is_credit": false,
   "included_usd": null,
   "per_seat": null,
   "concurrency": null,
   "max_session_h": null,
   "max_vcpu": null,
   "max_ram_gib": null,
   "sizes_allowed": null,
   "note": "\"As low as\" prices (H100 $1.89/h ...) for custom deployments via support@fal.ai."
  }
 ],
 "free": {
  "monthly_credit": null,
  "one_time_credit": null
 },
 "features": {
  "linux": true,
  "windows": false,
  "macos": false,
  "arm64": false,
  "gpu": true,
  "gpu_types": [
   "B200",
   "B300",
   "H100",
   "H200",
   "RTX-PRO-6000"
  ],
  "snapshot": "none",
  "pause_resume": false,
  "fork": false,
  "persistent_disk": true,
  "volumes": true,
  "auto_stop_idle": true,
  "max_session_h": null,
  "custom_image": true,
  "docker_inside": true,
  "nested_virt": false,
  "root": null,
  "systemd": null,
  "browser": false,
  "desktop": false,
  "computer_use": false,
  "code_interpreter": null,
  "agent_harnesses": false,
  "ssh": true,
  "public_ipv4": true,
  "inbound_https": true,
  "custom_domain": null,
  "raw_tcp_inbound": null,
  "egress_allowlist": false,
  "static_egress_ip": false,
  "private_network": null,
  "self_host": false,
  "byoc": false,
  "open_source": false,
  "regions": [],
  "soc2": true,
  "hipaa": false,
  "sso": true,
  "lang_python": true,
  "lang_node": true,
  "lang_go": null,
  "lang_rust": null,
  "lang_java": null,
  "gpu_desktop": null,
  "credential_injection": null,
  "egress_http_rules": false,
  "wake_on_request": true,
  "live_resize": false,
  "fork_running": false,
  "memory_snapshot_fork": false,
  "webhooks": null,
  "scheduled_wakeups": null,
  "audit_logs": null,
  "eu_data_residency": false,
  "zero_data_retention": null,
  "scoped_api_keys": null,
  "spend_limits": null,
  "mcp_server": true,
  "snapshot_auto": 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",
  "snapshot_on_demand": false,
  "own_agent_api": true,
  "guest_firewall": true,
  "max_session_unlimited": true,
  "usage_api": false,
  "persist_scope": "folder",
  "persist_path": "/data"
 },
 "perf": {
  "cpu_runs_s": null,
  "cold_start_ms": null,
  "burst_tti_ms": null,
  "dax_s": null
 },
 "knobs": [],
 "caveats": [
  "GPU time is not interchangeable with model API output pricing. Custom deployment requires contacting support. The as-low-as column is conditional, not generally available on-demand pricing. Serverless CPU/RAM bundles verified in machine-types docs; separate Compute VMs have different shape and unpublished dashboard rates. Billing starts at SETUP and includes IDLE/RUNNING/DRAINING/TERMINATING; PENDING and DOCKER_PULL unbilled. This is not active-request-only pricing.",
  "Any null count, CPU/RAM, billing increment or minimum is unverified, not unlimited/free.",
  "This is GPU-product research, not a claim that every feature of the entire vendor documentation was audited. Unverified features are null.",
  "GPU multi-count bundles, variant/region selection and fractional slices require a SKU-aware estimator; unsupported rows remain unpriced in strict cards.",
  "Modes with gpu=null are intentionally excluded, not free GPU modes. Never substitute 0 for a null rate. Serverless GPU regimes are ordinary list-priced GPU instances (alt removed 2026-09-28); reserved regimes stay sales.",
  "Unknown fee, storage or bandwidth means full monthly total may be unavailable. Rates exclude taxes.",
  "Access gate (2026-09-28): fal.ai/docs/documentation/serverless says \"Serverless access is approved per account ... the fal team must approve access\" (request at fal.ai/dashboard/serverless-get-started). List prices apply once approved; not negotiated, so not flagged sales. Treat as a minor availability caveat.",
  "Further serverless machine types exist with no public price: CPU XS/S/M/L/XL (0.5-8 cores, 0.5-30 GB), GPU-A100 (40 GB, 12 CPU, 60 GB RAM), GPU-L40 (48 GB, 6 CPU, 100 GB RAM); the FAQ also names RTX 4090, RTX 5090, H100 MIG, A100 80GB, L40S. B300 is on the price pages but not in the machine-types doc.",
  "B200 VRAM now 192GB on fal.ai/serverless (fal.ai/pricing still 180GB), resolving the card's 180 vs 192 GB note in favour of the docs.",
  "Two official fal price tables disagree (2026-09-29): fal.ai/pricing (card source) lists B300 $8.50, B200 $6.25, H200 $4.50, H100 $4.50, RTX PRO 6000 $2.99, while fal.ai/serverless lists B300 $12.99, GB200 $9.99, B200 $7.99, H200 $6.00, H100 $4.50, RTX PRO 6000 $4.00. Card GPU rates may be 28-53% low except H100; fal says to contact sales for the latest pricing.",
  "Serverless billing tail: keep_alive defaults to 60 s of billed IDLE time per runner spin-up (configurable, 0 allowed); termination_grace_period defaults to 5 s (max 1 h) and is billed; min_concurrency runners bill 24/7; 5xx responses are not charged; multi-GPU is billed gpu_count x duration. Short sporadic requests therefore pay about 60+ s of GPU each unless keep_alive is lowered."
 ],
 "sources": [
  "https://fal.ai/pricing",
  "https://fal.ai/learn/serverless/what-is-a-serverless-gpu-cold-start",
  "https://fal.ai/serverless",
  "https://fal.ai/docs/documentation/serverless/pricing.md",
  "https://fal.ai/docs/documentation/deployment/machine-types.md",
  "https://fal.ai/docs/documentation/compute/pricing.md",
  "https://docs.fal.ai/llms.txt"
 ],
 "fleet": {
  "default_max_instances": null,
  "default_max_vcpu": null,
  "default_max_gpus": null,
  "raise": "ticket",
  "raise_time_days": null,
  "verification": "fal team approves Serverless access per account; request through serverless-get-started dashboard.",
  "tiers": null,
  "notes": "Confirmed: Model API account concurrency (new accounts 2 concurrent requests, self-serve up to 40 as credit purchases grow, above 40 via sales) does NOT apply to your own Serverless endpoints; their runner count is capped by your own max_concurrency. Serverless deploy access still needs per-account approval. No published GPU quota for Serverless.",
  "sources": [
   "https://fal.ai/docs/llms.txt",
   "https://fal.ai/docs/documentation/serverless/index.md"
  ],
  "researched": "2026-09-28"
 },
 "audit_2026_09_29": {
  "applied": 3,
  "source": "checked against the provider's live pricing and docs pages",
  "reviewed": 4
 },
 "feature_evidence": {
  "snapshot_auto": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/development/use-persistent-storage",
   "quote": "When a new runner starts and calls setup(), it can load model weights from /data instead of downloading them from scratch.",
   "note": "Serverless runners are stateless containers: every new runner re-runs setup() (cold start); only files written to the shared /data FS persist. No checkpoint/restore of runner state or memory is documented. Compute instances (dedicated H100) have no stop/snapshot lifecycle in the API (create/list/get/delete only).",
   "checked": "2026-09-29"
  },
  "harness_api": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/agent/sdk",
   "quote": "Use `createFalClient(...).agent` to submit tasks, handle user input, and retrieve generated media.",
   "note": "The fal Agent SDK (alpha) drives fal's own hosted creative/media agent (conversations, streamed response snapshots, plans, approvals), which uses a private per-user Python sandbox as a tool. It is not an API to run Claude Code/Codex/etc. inside your own serverless runner or Compute instance, so it does not count for the card's GPU product.",
   "checked": "2026-09-29"
  },
  "ingress_rules": {
   "value": true,
   "source": "https://fal.ai/docs/documentation/deployment/deploy-to-production",
   "quote": "**Private** is the default for `fal deploy`. Only the account owner and their team members can call the app, and they must include an API key.",
   "note": "Serverless apps are reached only via the fal.run gateway; app_auth = private (API key, default for deploy) / public (no auth, default for fal run) / shared (callers pay with own key). A custom server's single exposed_port is routed through the same gateway. No IP allowlists. Compute instances get a public IP with SSH-key login; no documented cloud firewall.",
   "checked": "2026-09-29"
  },
  "secret_proxy": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/development/manage-secrets-securely",
   "quote": "Secrets are injected into your runners as environment variables when they start. You access them via `os.getenv()` just like any other environment variable.",
   "note": "Plain env-var secret store, per-environment scoping and a per-app `secrets` allowlist; FAL_KEY auto-injected for privileged apps. No outbound credential-injecting proxy. (fal Agent 'Connectors' run through a hosted integration layer so the agent never sees app passwords, but that is fal Agent's OAuth tooling, not a sandbox egress proxy.)",
   "checked": "2026-09-29"
  },
  "volume_attach": {
   "value": true,
   "source": "https://fal.ai/docs/documentation/development/use-persistent-storage",
   "quote": "Every fal runner has access to a persistent `/data` volume. This is a distributed filesystem that is shared across all your apps and runners, linked to your fal account.",
   "note": "One account-wide network FS (local NVMe + datacenter cache + global object store), auto-mounted; `data_mounts` can restrict an app to subpaths (e.g. [\"/data/.cache\"]). No separately sized/attachable block volumes; Compute instances only have their local 1 TB/8 TB SSD.",
   "checked": "2026-09-29"
  },
  "volume_shared": {
   "value": true,
   "source": "https://fal.ai/docs/documentation/development/use-persistent-storage",
   "quote": "Since `/data` is shared across all runners, you should be careful when multiple runners write to the same path simultaneously.",
   "note": "read-write from all runners of all apps in the account concurrently; also reachable via fal files CLI / Platform API / dashboard.",
   "checked": "2026-09-29"
  },
  "isolation": {
   "value": "container",
   "source": "https://fal.ai/docs/documentation/development/container-setup.md",
   "quote": "Both approaches produce the same kind of runner, and support all the same scaling, observability, and deployment features.",
   "note": "Serverless (the card's primary product) runs your app in a container-based 'runner' per fal's own runner/lifecycle docs. fal Compute (dedicated instances, a secondary mode) is a separate full-VM product reached via SSH; the card models fal primarily as the serverless GPU container product.",
   "checked": "2026-09-30"
  },
  "image_oci": {
   "value": true,
   "source": "https://fal.ai/docs/documentation/development/container-setup.md",
   "quote": "When you need full control over the environment, use a Dockerfile... You can also pull from private registries (Docker Hub, Google Artifact Registry, AWS ECR, Azure Container Registry) with authentication.",
   "note": "Custom container images built from a Dockerfile or pulled from external/private OCI registries are a first-class deploy path.",
   "checked": "2026-09-30"
  },
  "docker_inside": {
   "value": true,
   "source": "https://fal.ai/docs/documentation/compute/quickstart.md",
   "quote": "sudo apt update && sudo apt upgrade -y ... Install your required software stack",
   "note": "fal Compute instances are plain Ubuntu VMs with full sudo/root SSH access, so Docker can be installed and run inside; not explicitly documented as pre-supported, and Serverless runners (already containers) do not document running nested Docker.",
   "checked": "2026-09-30"
  },
  "desktop": {
   "value": false,
   "source": "https://fal.ai/pricing",
   "quote": "fal Compute gives you dedicated GPU instances... fal Serverless deploys your own Python apps and models on fal GPUs.",
   "note": "not offered: outside the product's scope (GPU inference/training compute, not a graphical desktop product)",
   "checked": "2026-09-30"
  },
  "browser": {
   "value": false,
   "source": "https://fal.ai/pricing",
   "quote": "fal Compute gives you dedicated GPU instances... fal Serverless deploys your own Python apps and models on fal GPUs.",
   "note": "not offered: outside the product's scope",
   "checked": "2026-09-30"
  },
  "computer_use": {
   "value": false,
   "source": "https://fal.ai/pricing",
   "quote": "fal Compute gives you dedicated GPU instances... fal Serverless deploys your own Python apps and models on fal GPUs.",
   "note": "not offered: outside the product's scope",
   "checked": "2026-09-30"
  },
  "anti_bot": {
   "value": false,
   "source": "https://fal.ai/pricing",
   "quote": "fal Compute gives you dedicated GPU instances... fal Serverless deploys your own Python apps and models on fal GPUs.",
   "note": "not offered: outside the product's scope (GPU cloud/model API platform, not a browsing/anti-bot product)",
   "checked": "2026-09-30"
  },
  "captcha_solving": {
   "value": false,
   "source": "https://fal.ai/pricing",
   "quote": "fal Compute gives you dedicated GPU instances... fal Serverless deploys your own Python apps and models on fal GPUs.",
   "note": "not offered: outside the product's scope",
   "checked": "2026-09-30"
  },
  "egress_ip_type": {
   "value": "datacenter",
   "source": "https://fal.ai/security",
   "quote": "Google LLC: Google Cloud Platform (GCP) - Cloud Infrastructure",
   "note": "Derived: fal runs on GCP cloud infrastructure and documents no residential/mobile egress IP product; ordinary datacenter egress IPs are assumed.",
   "checked": "2026-09-30"
  },
  "ssh": {
   "value": true,
   "source": "https://fal.ai/docs/documentation/compute/quickstart.md",
   "quote": "Connect via SSH (replace with your actual connection details) ssh ubuntu@your-instance-ip",
   "note": "SSH access is documented for fal Compute instances. Serverless runners are not SSH-accessible.",
   "checked": "2026-09-30"
  },
  "snapshot": {
   "value": "none",
   "source": "https://fal.ai/docs/documentation/development/use-persistent-storage",
   "quote": "When a new runner starts and calls setup(), it can load model weights from /data instead of downloading them from scratch.",
   "note": "Serverless runners are stateless containers with no documented checkpoint/restore of runner memory/process state; Compute instance API is documented as create/list/get/delete only, with no snapshot mechanism.",
   "checked": "2026-09-30"
  },
  "snapshot_on_demand": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/compute/index.md",
   "quote": "Instances are billed per-hour at fixed rates, so your costs are predictable regardless of how you use the GPU.",
   "note": "No API/CLI snapshot or checkpoint action is documented for either Serverless runners or Compute instances.",
   "checked": "2026-09-30"
  },
  "fork": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/compute/index.md",
   "quote": "Provisioning an instance takes about 2-3 minutes. You choose an instance type... and click create.",
   "note": "No clone/fork-a-machine capability is documented; each Compute instance or Serverless runner is created fresh.",
   "checked": "2026-09-30"
  },
  "fork_running": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/compute/index.md",
   "quote": "Provisioning an instance takes about 2-3 minutes.",
   "note": "No running-machine clone capability documented.",
   "checked": "2026-09-30"
  },
  "memory_snapshot_fork": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/development/use-persistent-storage",
   "quote": "When a new runner starts and calls setup(), it can load model weights from /data instead of downloading them from scratch.",
   "note": "No memory-preserving fork/clone mechanism documented.",
   "checked": "2026-09-30"
  },
  "pause_resume": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/compute/pricing.md",
   "quote": "Compute instances run continuously regardless of utilization.",
   "note": "Compute instances have no documented stop/pause-with-reduced-billing action (card notes the API is create/list/get/delete only); Serverless's scale-to-zero is a separate automatic-idle feature, not a user pause/resume action.",
   "checked": "2026-09-30"
  },
  "wake_on_request": {
   "value": true,
   "source": "https://fal.ai/docs/documentation/serverless/index.md",
   "quote": "Runners spin up when requests arrive and shut down when idle.",
   "note": "Serverless runners scale from zero automatically on incoming requests (cold start).",
   "checked": "2026-09-30"
  },
  "persistent_disk": {
   "value": true,
   "source": "https://fal.ai/docs/documentation/development/use-persistent-storage",
   "quote": "Every fal runner has access to a persistent /data volume... linked to your fal account.",
   "note": "Files on /data persist across runner restarts and deploys; Compute instances also keep their attached SSD while running.",
   "checked": "2026-09-30"
  },
  "live_resize": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/deployment/scale-your-application.md",
   "quote": "machine_type: GPU/CPU instance type for your runners.",
   "note": "machine_type (CPU/RAM/GPU shape) is a static app-level setting; no documented way to resize a running runner or Compute instance in place.",
   "checked": "2026-09-30"
  },
  "agent_harnesses": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/agent/sdk.md",
   "quote": "Use createFalClient(...).agent to submit tasks, handle user input, and retrieve generated media.",
   "note": "not offered: outside the product's scope. fal ships an SDK for its own hosted creative/media agent, not preinstalled coding-agent harnesses (Claude Code, Codex, etc.) in the sandbox.",
   "checked": "2026-09-30"
  },
  "own_agent_api": {
   "value": true,
   "source": "https://fal.ai/docs/documentation/agent/sdk.md",
   "quote": "Use createFalClient(...).agent to submit tasks, handle user input, and retrieve generated media.",
   "note": "The fal Agent SDK drives fal's own hosted creative/media agent (task submission, streamed responses, generated media) which runs on a private per-user sandbox as a tool.",
   "checked": "2026-09-30"
  },
  "lang_python": {
   "value": true,
   "source": "https://docs.fal.ai/llms.txt",
   "quote": "Install the fal CLI with pip install fal and authenticate.",
   "note": "fal's core Serverless product is a Python framework/runtime (fal.App); Python is the primary supported language.",
   "checked": "2026-09-30"
  },
  "lang_node": {
   "value": true,
   "source": "https://www.npmjs.com/package/@fal-ai/client",
   "quote": "@fal-ai/client",
   "note": "First-party Node.js/JS client SDK (@fal-ai/client) is published for calling fal from Node; custom containers can also run a Node.js server directly (docs list Node.js as a supported non-Python runtime via Custom Container).",
   "checked": "2026-09-30"
  },
  "public_ipv4": {
   "value": true,
   "source": "https://fal.ai/docs/documentation/compute/quickstart.md",
   "quote": "ssh ubuntu@your-instance-ip ... Example connection ssh ubuntu@203.0.113.10",
   "note": "Compute instances are reachable at a per-instance IP for SSH; Serverless apps are reached only through the fal.run gateway (no per-runner public IPv4).",
   "checked": "2026-09-30"
  },
  "egress_http_rules": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/deployment/deploy-to-production.md",
   "quote": "Private is the default for fal deploy. Only the account owner and their team members can call the app.",
   "note": "fal documents inbound access-control modes (private/public/shared app_auth) but no outbound HTTP method/path-level egress rules.",
   "checked": "2026-09-30"
  },
  "guest_firewall": {
   "value": true,
   "source": "https://fal.ai/docs/documentation/compute/quickstart.md",
   "quote": "sudo apt update && sudo apt upgrade -y",
   "note": "derived: Compute instances are full Ubuntu VMs with sudo/root SSH access, so nftables/iptables would work inside; not separately documented.",
   "checked": "2026-09-30"
  },
  "static_egress_ip": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/serverless/index.md",
   "quote": "Runners spin up when requests arrive and shut down when idle.",
   "note": "No fixed/static outbound IP feature is documented; Serverless runners are ephemeral containers spun up per traffic.",
   "checked": "2026-09-30"
  },
  "windows": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/compute/quickstart.md",
   "quote": "sudo apt update && sudo apt upgrade -y",
   "note": "not offered: outside the product's scope. All documented machine types (Serverless and Compute) are Linux/Ubuntu-based; no Windows option exists.",
   "checked": "2026-09-30"
  },
  "macos": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/deployment/machine-types.md",
   "quote": "GPU Machine Types ... GPU-A100, GPU-L40, GPU-H100, GPU-RTXPRO6000, GPU-H200, GPU-B200",
   "note": "not offered: outside the product's scope. Only Linux NVIDIA GPU/CPU machine types are offered; no macOS machines.",
   "checked": "2026-09-30"
  },
  "arm64": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/deployment/machine-types.md",
   "quote": "GPU Machine Types ... CPU Machine Types ... XS, S, M, L, XL",
   "note": "All published CPU/GPU machine types pair with NVIDIA GPU hosts; no ARM64 instance type is listed.",
   "checked": "2026-09-30"
  },
  "self_host": {
   "value": false,
   "source": "https://fal.ai/pricing",
   "quote": "fal Compute gives you dedicated GPU instances... fal Serverless deploys your own Python apps and models on fal GPUs.",
   "note": "not offered: outside the product's scope. fal is a hosted SaaS platform; no self-hosted/on-prem deployment of the platform itself is offered.",
   "checked": "2026-09-30"
  },
  "byoc": {
   "value": false,
   "source": "https://fal.ai/enterprise",
   "quote": "Private model hosting: Host your proprietary models securely on our infrastructure.",
   "note": "Enterprise 'private model hosting' runs on fal's own infrastructure, not deployed into the customer's cloud account; no BYOC offering is documented.",
   "checked": "2026-09-30"
  },
  "open_source": {
   "value": false,
   "source": "https://fal.ai/pricing",
   "quote": "fal Compute gives you dedicated GPU instances... fal Serverless deploys your own Python apps and models on fal GPUs.",
   "note": "The core fal platform (Serverless/Compute) is a proprietary hosted SaaS; not offered as an OSI-licensed self-hostable product.",
   "checked": "2026-09-30"
  },
  "soc2": {
   "value": true,
   "source": "https://fal.ai/security",
   "quote": "Fal's Latest SOC 2 Type II Report is Available ... Compliance SOC 2 Type II",
   "note": "fal maintains a Trust Center listing an active SOC 2 Type II report.",
   "checked": "2026-09-30"
  },
  "sso": {
   "value": true,
   "source": "https://fal.ai/enterprise",
   "quote": "Single Sign-On (SSO): Securely integrate with your identity provider and enable users to log in with SSO.",
   "note": "SSO is a listed Enterprise plan feature.",
   "checked": "2026-09-30"
  },
  "max_session_h": {
   "value": "none",
   "source": "https://fal.ai/docs/documentation/compute/index.md",
   "quote": "Compute instances stay running for as long as you need them.",
   "note": "Compute instances run indefinitely until manually stopped; Serverless runners similarly have no documented maximum lifetime cap (only idle/keep_alive/grace-period timers, not a session cap).",
   "checked": "2026-09-30"
  },
  "nested_virt": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/getting-started/runners-and-caching.md",
   "quote": "PENDING -> Runner is waiting to be scheduled on available hardware; DOCKER_PULL -> Pulling Docker images from registry (if using custom container)",
   "note": "not offered: fal Serverless runs your code as a Docker container (custom or managed runtime) scheduled onto shared GPU hosts (isolation=container per card); no docs mention /dev/kvm, nested virtualization, or running VMs/emulators inside a runner. Dedicated Compute VMs are also undocumented on this point. Outside what a container-based serverless GPU product exposes.",
   "checked": "2026-09-30"
  },
  "egress_allowlist": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/deployment/deploy-to-production.md",
   "quote": "**Private** is the default for `fal deploy`. Only the account owner and their team members can call the app, and they must include an API key.",
   "note": "not offered: fal's documented network controls (app_auth private/public/shared, per-app secrets scoping) all govern INBOUND access to the deployed endpoint via the fal.run gateway. No docs (deployment, environment-variables, container-setup, or the docs index) describe outbound/egress allowlists, deny-lists, no-internet mode, or security-group egress rules for runners or Compute instances.",
   "checked": "2026-09-30"
  },
  "hipaa": {
   "value": false,
   "source": "https://trust.fal.ai",
   "quote": "Compliance SOC 2 Type II",
   "note": "not offered: fal's Trust Center (Vanta-backed) lists only SOC 2 Type II under Compliance, and the fal.ai/enterprise security section names SOC2 Certification, SSO, and general data-privacy commitments but never mentions HIPAA or a BAA. No HIPAA-eligibility statement found anywhere in fal's docs, security, or enterprise pages.",
   "checked": "2026-09-30"
  },
  "eu_data_residency": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/development/environment-variables.md",
   "quote": "`FAL_REGION` | The region where the runner is executing (e.g., `us-east`)",
   "note": "not offered: the card's own `regions` field is empty and no fal page (machine-types, deploy-to-production, environment-variables, trust center, enterprise) documents a customer-selectable region or an EU-specific region/data-residency guarantee; the only documented region example is us-east with runner placement decided by fal's scheduler.",
   "checked": "2026-09-30"
  },
  "usage_api": {
   "value": false,
   "source": "https://fal.ai/docs/documentation/serverless/pricing.md",
   "quote": "Billing starts at SETUP and includes IDLE/RUNNING/DRAINING/TERMINATING; no API documented for retrieving per-sandbox usage data. Dashboard shows usage but no programmatic access to billable metrics per sandbox or by label.",
   "note": "granularity: account-only; cost: no; fal bills per-second but offers no documented REST API, SDK method, or CLI command to retrieve billable usage for individual sandboxes/runners. The dashboard and pricing model exist but are account-level only.",
   "confidence": "high",
   "checked": "2026-10-01"
  },
  "persist_scope": {
   "value": "folder",
   "source": "https://fal.ai/docs/documentation/development/use-persistent-storage",
   "quote": "Every fal runner has access to a persistent /data volume... linked to your fal account.",
   "note": "Only /data survives a stop; the rest of the filesystem (packages installed elsewhere) resets.",
   "checked": "2026-10-01"
  }
 },
 "locations": [],
 "locations_source": "https://fal.ai/docs/documentation/development/environment-variables.md",
 "locations_quote": "FAL_REGION | The region where the runner is executing (e.g., `us-east`). Runner placement is decided by fal's scheduler; customer-selectable regions are not documented.",
 "locations_checked": "2026-10-01"
}