{
 "features_resolved": {
  "snapshot_any": false,
  "snapshot_mem": false,
  "fork": false,
  "pause_resume": false,
  "persistent_disk": false,
  "whole_disk": false,
  "persist_scope": null,
  "volumes": false,
  "auto_stop_idle": true,
  "long_sessions": true,
  "custom_image": true,
  "image_snapshot": false,
  "image_oci": true,
  "vm_isolation": false,
  "docker_inside": false,
  "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": false,
  "public_ipv4": false,
  "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": true,
  "byoc": false,
  "open_source": true,
  "soc2": null,
  "hipaa": null,
  "sso": false,
  "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": true,
  "scheduled_wakeups": null,
  "audit_logs": true,
  "eu_data_residency": false,
  "zero_data_retention": null,
  "scoped_api_keys": null,
  "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": false,
  "guest_firewall": false,
  "secret_proxy": false,
  "secret_proxy_any": false,
  "volume_attach": false,
  "volume_shared": false,
  "sidecar": null,
  "fuse": null,
  "kernel_features": null,
  "burst": null
 },
 "company": {
  "id": "replicate",
  "checked": "2026-09-30",
  "company": "Replicate, Inc.",
  "ownership": {
   "value": "acquired",
   "parent": "Cloudflare",
   "source": "https://replicate.com/blog/replicate-cloudflare",
   "quote": "Big news: We’re joining Cloudflare. Replicate’s going to carry on as a distinct brand, and all that’ll happen is that it’s going to get way better."
  },
  "stage": {
   "value": "subsidiary",
   "source": "https://replicate.com/blog/replicate-cloudflare",
   "quote": "Big news: We’re joining Cloudflare.",
   "confidence": "high",
   "note": "Replicate was acquired by / joined Cloudflare (announced 2025-11-17) and continues as a distinct brand; its last independent priced round was a $40M Series B (Dec 2023), so pre-acquisition stage was series-b."
  },
  "accelerators": [
   {
    "name": "Y Combinator",
    "batch": "Winter 2020",
    "source": "https://www.ycombinator.com/companies/replicate",
    "quote": "Replicate Run machine learning models in the cloud WINTER 2020 ACQUIRED ... Founded: 2019 Batch: Winter 2020 Team Size: 36 Status: Acquired",
    "confidence": "high",
    "via": "research"
   },
   {
    "name": "AI Grant",
    "batch": "1",
    "source": "https://aigrant.com/",
    "quote": "listed in the AI Grant portfolio as \"Replicate\"",
    "confidence": "high",
    "via": "portfolio-domain"
   }
  ],
  "total_funding_usd": {
   "value": 57800000,
   "status": "estimated",
   "basis": "Aggregator figure (~$57.8M per AI-summarized Crunchbase-style snippet; growjo.com separately estimates total funding at $58M) matching sum of seed (undisclosed, part of $17.8M pre-Series-A total) + $12.5M Series A (Feb 2023) + $40M Series B (Dec 2023)",
   "source": "https://techcrunch.com/2023/02/21/replicate-wants-to-take-the-pain-out-of-running-and-hosting-ml-models/",
   "quote": "Replicate, a startup that runs machine learning models in the cloud, today launched out of stealth with $17.8 million in venture capital backing; $12.5 million of the total came from a Series A led by Andreessen Horowitz with participation from Y Combinator, Sequoia and angel investors"
  },
  "last_round": {
   "type": "series-b",
   "amount_usd": 40000000,
   "date": "2023-12",
   "leads": [
    "Andreessen Horowitz (a16z)"
   ],
   "status": "confirmed",
   "source": "https://replicate.com/blog/series-b",
   "quote": "we’ve raised a $40 million Series B led by a16z ... It’s led by a16z, and joined by NVentures (NVIDIA), Heavybit, Sequoia, and Y Combinator."
  },
  "investors": [
   "Andreessen Horowitz (a16z)",
   "Sequoia Capital",
   "Y Combinator",
   "NVentures (NVIDIA)",
   "Heavybit",
   "Dylan Field (Figma CEO, angel)",
   "Guillermo Rauch (Vercel, angel)"
  ],
  "portfolio_listings": [
   {
    "source": "AI Grant",
    "company": "Replicate",
    "batch": "1",
    "how": "domain"
   },
   {
    "source": "Andreessen Horowitz",
    "company": "Replicate",
    "batch": null,
    "how": "exact-name"
   },
   {
    "source": "Heavybit",
    "company": "Replicate",
    "batch": null,
    "how": "domain"
   }
  ],
  "revenue": {
   "value_usd": 5300000,
   "kind": "annual revenue",
   "as_of": "2024",
   "status": "estimated",
   "basis": "GetLatka estimate; growjo.com separately estimates current annual revenue at $6.3M",
   "source": "https://getlatka.com/companies/replicate",
   "quote": "Replicate generated an estimated $5.3M in annual revenue in 2024. Source: GetLatka estimate"
  },
  "employees": {
   "value": 36,
   "source": "https://www.ycombinator.com/companies/replicate"
  },
  "note": "Independent startup (YC W20, founded 2018-2019 by Ben Firshman [ex-Docker] and Andreas Jansson [ex-Spotify]) that raised seed + $12.5M Series A (Feb 2023, a16z-led, $17.8M total at launch) then a $40M Series B (Dec 2023, a16z-led). On 2025-11-17 Replicate announced it is joining Cloudflare (acquisition/acqui-integration); it continues to operate as a distinct brand with an unchanged API and no pricing change observed as of 2026-09-29. Revenue figures are third-party estimates (GetLatka $5.3M for 2024, growjo ~$6.3M current), not company-confirmed."
 },
 "id": "replicate",
 "name": "Replicate",
 "url": "https://replicate.com/pricing",
 "category": "gpu-cloud",
 "isolation": "container",
 "modes": [
  {
   "key": "private-cpu",
   "label": "Private Cog CPU model",
   "flags": [
    "alt"
   ],
   "os": [
    "linux"
   ],
   "regions": null,
   "gpu_only": false,
   "features": {},
   "pricing": "sizes",
   "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": [
    {
     "name": "cpu-small",
     "vcpu": 1,
     "ram_gib": 2,
     "disk_gib": null,
     "hour": 0.09,
     "month_cap": null
    },
    {
     "name": "cpu",
     "vcpu": 4,
     "ram_gib": 8,
     "disk_gib": null,
     "hour": 0.36,
     "month_cap": 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": null,
   "note": "Private models and deployments bill setup + idle + active instance time, per second. After traffic stops an instance idles 'for a few minutes' before scaling to min_instances (0 by default), and that idle tail is billed. Failed and canceled runs are billed for private models and deployments. Same applies to all serverless-* GPU modes (unless running a public model or a fast-booting fine-tune, which bill active time only).",
   "cpu_class": "dedicated"
  },
  {
   "key": "serverless-t4-0",
   "label": "T4, serverless (×1)",
   "flags": [
    "alt"
   ],
   "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": 4,
   "max_ram_gib": 16,
   "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": {
    "T4": 0.81
   },
   "note": "Source: https://replicate.com/pricing [grafted from v4-gpu card 2026-09-28]",
   "cpu_class": "dedicated"
  },
  {
   "key": "serverless-a100-80gb-1",
   "label": "A100-80GB, serverless (×1)",
   "flags": [
    "alt"
   ],
   "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": 10,
   "max_ram_gib": 144,
   "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": {
    "A100-80GB": 5.04
   },
   "note": "Source: https://replicate.com/pricing [grafted from v4-gpu card 2026-09-28]",
   "cpu_class": "dedicated"
  },
  {
   "key": "serverless-h100-2",
   "label": "H100, serverless (×1)",
   "flags": [
    "alt"
   ],
   "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": 13,
   "max_ram_gib": 144,
   "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": 5.49
   },
   "note": "Source: https://replicate.com/pricing [grafted from v4-gpu card 2026-09-28]",
   "cpu_class": "dedicated"
  },
  {
   "key": "serverless-l40s-3",
   "label": "L40S, serverless (×1)",
   "flags": [
    "alt"
   ],
   "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": 10,
   "max_ram_gib": 65,
   "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": {
    "L40S": 3.51
   },
   "note": "Source: https://replicate.com/pricing [grafted from v4-gpu card 2026-09-28]",
   "cpu_class": "dedicated"
  },
  {
   "key": "serverless-a100-80gb-2x",
   "label": "2x A100-80GB, serverless",
   "flags": [
    "alt"
   ],
   "pricing": "resource",
   "gpu_only": true,
   "gpu_min_count": 2,
   "vcpu_h": 0,
   "ram_gib_h": 0,
   "max_vcpu": 20,
   "max_ram_gib": 288,
   "granularity_s": 1,
   "gpu": {
    "A100-80GB": 5.04
   },
   "note": "gpu-a100-large-2x $0.002800/sec = $10.08/h for the node",
   "cpu_class": "dedicated"
  },
  {
   "key": "serverless-l40s-2x",
   "label": "2x L40S, serverless",
   "flags": [
    "alt"
   ],
   "pricing": "resource",
   "gpu_only": true,
   "gpu_min_count": 2,
   "vcpu_h": 0,
   "ram_gib_h": 0,
   "max_vcpu": 20,
   "max_ram_gib": 144,
   "granularity_s": 1,
   "gpu": {
    "L40S": 3.51
   },
   "note": "gpu-l40s-2x $0.001950/sec = $7.02/h for the node",
   "cpu_class": "dedicated"
  },
  {
   "key": "committed-multi-gpu",
   "label": "Multi-GPU / H200 (committed spend)",
   "flags": [
    "alt",
    "sales"
   ],
   "pricing": "resource",
   "gpu_only": true,
   "vcpu_h": 0,
   "ram_gib_h": 0,
   "granularity_s": 1,
   "gpu": {
    "A100-80GB": 5.04,
    "H100": 5.49,
    "H200": 5.49,
    "L40S": 3.51
   },
   "note": "Listed per node: A100 4x $20.16/h, 8x $40.32/h; H100 2x $10.98, 4x $21.96, 8x $43.92; H200 1x $5.49, 2x $10.98, 4x $21.96, 8x $43.92; L40S 4x $14.04, 8x $28.08. Per-GPU rate equals the 1x rate. Capacity needs a committed-spend contract (commitment size not published).",
   "cpu_class": "dedicated"
  }
 ],
 "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": "Usage",
   "fee": 0,
   "flags": [],
   "fee_is_credit": false,
   "included_usd": 0,
   "per_seat": null,
   "concurrency": null,
   "max_session_h": null,
   "max_vcpu": null,
   "max_ram_gib": null,
   "sizes_allowed": null,
   "note": "Prepaid credit required (\"To use Replicate, you must first purchase credit upfront.\", https://replicate.com/docs/topics/billing/prepaid-credit, checked 2026-10-14; the page shows no account-date carve-out): usage is deducted from the balance; purchased credit expires 1 year after purchase and is non-refundable; auto-reload optional (min threshold $5, min reload balance $15). At $0 balance new work is blocked and running infrastructure is shut down. A prediction that runs past the balance is charged to the default payment method at month end."
  }
 ],
 "free": {
  "monthly_credit": null,
  "one_time_credit": null
 },
 "features": {
  "linux": true,
  "windows": false,
  "macos": false,
  "arm64": false,
  "gpu": true,
  "snapshot": "none",
  "pause_resume": false,
  "fork": false,
  "persistent_disk": false,
  "volumes": null,
  "auto_stop_idle": true,
  "max_session_h": null,
  "custom_image": true,
  "docker_inside": false,
  "nested_virt": false,
  "root": null,
  "systemd": null,
  "browser": false,
  "desktop": false,
  "computer_use": false,
  "code_interpreter": null,
  "agent_harnesses": false,
  "ssh": false,
  "public_ipv4": false,
  "inbound_https": true,
  "custom_domain": null,
  "raw_tcp_inbound": null,
  "egress_allowlist": false,
  "static_egress_ip": false,
  "private_network": null,
  "self_host": true,
  "byoc": false,
  "open_source": true,
  "soc2": null,
  "hipaa": null,
  "sso": false,
  "gpu_types": [
   "T4",
   "L40S",
   "A100-80GB",
   "H100",
   "H200"
  ],
  "regions": null,
  "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": true,
  "scheduled_wakeups": null,
  "audit_logs": true,
  "eu_data_residency": false,
  "zero_data_retention": null,
  "scoped_api_keys": null,
  "spend_limits": true,
  "mcp_server": true,
  "secret_proxy": false,
  "isolation": "container",
  "image_oci": true,
  "anti_bot": false,
  "captcha_solving": false,
  "egress_ip_type": "datacenter",
  "snapshot_auto": false,
  "snapshot_on_demand": false,
  "volume_attach": false,
  "volume_shared": false,
  "harness_api": false,
  "own_agent_api": false,
  "ingress_rules": false,
  "guest_firewall": false,
  "max_session_unlimited": true,
  "usage_api": false
 },
 "perf": {
  "cpu_runs_s": null,
  "cold_start_ms": null,
  "burst_tti_ms": null,
  "dax_s": null
 },
 "caveats": [
  "Public inference charges only processing (some by output/token); private custom Cog models charge setup + idle + active dedicated instance time. Fast-booting fine-tunes are special active-only exceptions.",
  "Private CPU model deployment is an alt to interactive VM. GPUs include their hosts; do not double-count host CPU/RAM. Several multi-GPU and H200 SKUs require committed-spend contracts.",
  "Null values mean unknown, not free/unlimited. No model-token cost, tax or unverified ancillary charge included.",
  "Native browser/device/task/credit prices with unknown CPU/RAM are deliberately non-priceable. Check regime before interpreting any partial resource estimate.",
  "GPU modes grafted from the v4-gpu research card (per-GPU rates; exact rows in research/gpu.json).",
  "Replicate joined Cloudflare (announced 2025-11-17) and continues as a distinct brand with an unchanged API; no pricing change observed on the live pricing page as of 2026-09-29.",
  "A single prediction times out after 30 minutes (\"Predictions time out after running for 30 minutes.\", https://replicate.com/docs/topics/predictions/lifecycle); longer only via support. Instances have no session cap, so max_session_h stays unlimited, but any one job over 30 minutes needs a support request.",
  "API prediction inputs, outputs, files and logs are deleted after an hour by default (https://replicate.com/docs/topics/predictions/data-retention); web-created predictions are kept indefinitely. Not a zero-retention guarantee."
 ],
 "sources": [
  "https://www.ycombinator.com/companies/replicate",
  "https://replicate.com/pricing",
  "https://replicate.com/docs/guides/deploy-a-custom-model",
  "https://replicate.com/",
  "https://replicate.com/docs",
  "https://replicate.com/docs/topics/predictions/lifecycle",
  "https://replicate.com/docs/topics/billing/prepaid-credit",
  "https://replicate.com/docs/topics/predictions/data-retention"
 ],
 "fleet": {
  "default_max_instances": null,
  "default_max_vcpu": null,
  "default_max_gpus": null,
  "raise": "ticket",
  "raise_time_days": null,
  "verification": null,
  "tiers": null,
  "notes": "Deployments expose min/max instances for autoscaling; API documents rate limits and contact for higher rates, not a fixed new-account GPU/VM allowance. Request-rate increase is distinct from fleet capacity; numeric aggregate quota and review SLA not established. Create-prediction API limited to 600 requests/min (other endpoints 3,000/min); accounts with granted credit and no payment method are limited to 1 req/s, max 6/min; stronger throttling as prepaid credit runs low. Higher limits via support.",
  "sources": [
   "https://replicate.com/docs/topics/deployments",
   "https://replicate.com/docs/topics/predictions/rate-limits"
  ],
  "researched": "2026-09-28"
 },
 "audit_2026_09_29": {
  "applied": 9,
  "source": "checked against the provider's live pricing and docs pages",
  "reviewed": 3
 },
 "feature_evidence": {
  "secret_proxy": {
   "value": false,
   "source": "https://replicate.com/docs/topics/predictions/secrets",
   "quote": "Once the model receives your secret, it’s removed from Replicate’s systems",
   "note": "Secrets are Cog `Secret` typed prediction inputs passed into the model container (read with get_secret_value()); they are only redacted from logs/metadata/API responses. The secret does enter the container. No credential-injecting proxy.",
   "checked": "2026-09-29"
  },
  "isolation": {
   "value": "container",
   "source": "https://cog.run/",
   "quote": "Cog is an open-source tool that lets you package machine learning models in a standard, production-ready container.",
   "note": "Models run as Docker containers (Cog); no VM/microVM boundary is documented per customer.",
   "checked": "2026-09-30"
  },
  "image_oci": {
   "value": true,
   "source": "https://cog.run/",
   "quote": "Or, build a Docker image for deployment: $ cog build -t my-classification-model",
   "note": "You define your model with cog.yaml/predict.py, Cog builds a standard Docker image which is pushed and run on Replicate.",
   "checked": "2026-09-30"
  },
  "docker_inside": {
   "value": false,
   "source": "https://cog.run/",
   "quote": "Cog is an open-source tool that lets you package machine learning models in a standard, production-ready container.",
   "note": "not offered: the hosted platform runs your single Cog container image; no documented support for running Docker/containers nested inside that container.",
   "checked": "2026-09-30"
  },
  "nested_virt": {
   "value": false,
   "source": "https://replicate.com/pricing",
   "quote": "Run your model on Replicate",
   "note": "not offered: outside the product's scope (GPU model-inference containers, not a VM/emulation platform; no /dev/kvm documented).",
   "checked": "2026-09-30"
  },
  "desktop": {
   "value": false,
   "source": "https://replicate.com/docs",
   "quote": "Run and fine-tune open-source models. Deploy custom models at scale. All with one line of code.",
   "note": "not offered: outside the product's scope (API-driven model hosting, no graphical desktop).",
   "checked": "2026-09-30"
  },
  "browser": {
   "value": false,
   "source": "https://replicate.com/docs",
   "quote": "Run and fine-tune open-source models. Deploy custom models at scale. All with one line of code.",
   "note": "not offered: outside the product's scope; no managed/real browser is provided.",
   "checked": "2026-09-30"
  },
  "computer_use": {
   "value": false,
   "source": "https://replicate.com/docs",
   "quote": "Run and fine-tune open-source models. Deploy custom models at scale. All with one line of code.",
   "note": "not offered: outside the product's scope; Replicate hosts model inference, not a screenshot/click desktop-driving API.",
   "checked": "2026-09-30"
  },
  "anti_bot": {
   "value": false,
   "source": "https://replicate.com/docs",
   "quote": "Run and fine-tune open-source models. Deploy custom models at scale. All with one line of code.",
   "note": "not offered: outside the product's scope (GPU model-inference hosting, not a browser/scraping product).",
   "checked": "2026-09-30"
  },
  "captcha_solving": {
   "value": false,
   "source": "https://replicate.com/docs",
   "quote": "Run and fine-tune open-source models. Deploy custom models at scale. All with one line of code.",
   "note": "not offered: outside the product's scope.",
   "checked": "2026-09-30"
  },
  "egress_ip_type": {
   "value": "datacenter",
   "source": "https://replicate.com/docs/topics/site-policy/subprocessors",
   "quote": "AWS Cloud Hosted Infrastructure and Data Hosting United States",
   "note": "Runs on hyperscaler/GPU-cloud datacenter infrastructure (AWS, CoreWeave, etc.); no residential/mobile egress documented.",
   "checked": "2026-09-30"
  },
  "ssh": {
   "value": false,
   "source": "https://replicate.com/docs/guides/deploy-a-custom-model",
   "quote": "Replicate lets you build and scale AI products with your own custom models. You write the code, we handle the infrastructure orchestration and scaling.",
   "note": "not offered: models are invoked only via the HTTP/client-library prediction API; no SSH access to the running container is documented.",
   "checked": "2026-09-30"
  },
  "snapshot": {
   "value": "none",
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Auto-scaling: Scale from zero to hundreds of instances based on traffic",
   "note": "No filesystem or memory snapshot/checkpoint feature is documented; instances scale to zero and cold-start fresh from the image.",
   "checked": "2026-09-30"
  },
  "snapshot_auto": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Auto-scaling: Scale from zero to hundreds of instances based on traffic",
   "note": "not offered: no automatic state snapshot/restore is documented.",
   "checked": "2026-09-30"
  },
  "snapshot_on_demand": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Auto-scaling: Scale from zero to hundreds of instances based on traffic",
   "note": "not offered: no on-demand snapshot/checkpoint API is documented.",
   "checked": "2026-09-30"
  },
  "fork": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Deployments give you production-grade control over your model's infrastructure and provide private, dedicated API endpoints.",
   "note": "not offered: no clone-a-running-machine feature; you deploy the same image to more replicas via autoscaling instead.",
   "checked": "2026-09-30"
  },
  "fork_running": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Deployments give you production-grade control over your model's infrastructure and provide private, dedicated API endpoints.",
   "note": "not offered.",
   "checked": "2026-09-30"
  },
  "memory_snapshot_fork": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Deployments give you production-grade control over your model's infrastructure and provide private, dedicated API endpoints.",
   "note": "not offered.",
   "checked": "2026-09-30"
  },
  "pause_resume": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Auto-scaling: Scale from zero to hundreds of instances based on traffic",
   "note": "not offered as a distinct pause/resume API; only scale-to-zero/autoscaling is documented.",
   "checked": "2026-09-30"
  },
  "wake_on_request": {
   "value": true,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Auto-scaling: Scale from zero to hundreds of instances based on traffic",
   "note": "Deployments/serverless models scale from zero automatically when a prediction request arrives (incurring a cold start).",
   "checked": "2026-09-30"
  },
  "persistent_disk": {
   "value": false,
   "source": "https://replicate.com/docs/guides/deploy-a-custom-model",
   "quote": "Run arbitrary code and weights on high-end GPUs (A100s, H100s, B200s, and more) via API",
   "note": "not offered: model containers are stateless inference workers; no documented persistent disk that survives stop/start outside the built image/weights.",
   "checked": "2026-09-30"
  },
  "volume_attach": {
   "value": false,
   "source": "https://replicate.com/docs/guides/deploy-a-custom-model",
   "quote": "Run arbitrary code and weights on high-end GPUs (A100s, H100s, B200s, and more) via API",
   "note": "not offered: no attachable persistent volume feature documented.",
   "checked": "2026-09-30"
  },
  "volume_shared": {
   "value": false,
   "source": "https://replicate.com/docs/guides/deploy-a-custom-model",
   "quote": "Run arbitrary code and weights on high-end GPUs (A100s, H100s, B200s, and more) via API",
   "note": "not offered.",
   "checked": "2026-09-30"
  },
  "live_resize": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Hardware flexibility: Choose from multiple GPU architectures including NVIDIA A100s, H100s, T4s, and more. Switch hardware types without changing your code",
   "note": "Hardware type is switched by redeploying, not resizing a live running instance; no live in-place resize documented.",
   "checked": "2026-09-30"
  },
  "agent_harnesses": {
   "value": false,
   "source": "https://replicate.com/docs",
   "quote": "Run and fine-tune open-source models. Deploy custom models at scale. All with one line of code.",
   "note": "not offered: outside the product's scope (a model-hosting API, not a coding-agent sandbox with preinstalled harnesses).",
   "checked": "2026-09-30"
  },
  "harness_api": {
   "value": false,
   "source": "https://replicate.com/docs",
   "quote": "Run and fine-tune open-source models. Deploy custom models at scale. All with one line of code.",
   "note": "not offered: outside the product's scope.",
   "checked": "2026-09-30"
  },
  "own_agent_api": {
   "value": false,
   "source": "https://replicate.com/docs",
   "quote": "Run and fine-tune open-source models. Deploy custom models at scale. All with one line of code.",
   "note": "not offered: Replicate hosts third-party/custom models via API; it does not offer its own browser/computer-use/research task agent.",
   "checked": "2026-09-30"
  },
  "lang_python": {
   "value": true,
   "source": "https://replicate.com/docs",
   "quote": "Run a model from Python",
   "note": "Official Python client library and Cog (Python-based model packaging) are first-party.",
   "checked": "2026-09-30"
  },
  "lang_node": {
   "value": true,
   "source": "https://replicate.com/docs",
   "quote": "Run a model from Node.js",
   "note": "Official Node.js (replicate-javascript) client library documented.",
   "checked": "2026-09-30"
  },
  "public_ipv4": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Deployments give you production-grade control over your model's infrastructure and provide private, dedicated API endpoints.",
   "note": "not offered: access is via Replicate's HTTPS API/endpoints, not a per-machine public IPv4 address.",
   "checked": "2026-09-30"
  },
  "ingress_rules": {
   "value": false,
   "source": "https://replicate.com/docs/topics/security/api-tokens",
   "quote": "API tokens are used to authenticate your requests to the Replicate API.",
   "note": "Access control is via API token auth, not user-configurable IP allowlists/security groups on exposed ports.",
   "checked": "2026-09-30"
  },
  "egress_allowlist": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Deployments give you production-grade control over your model's infrastructure and provide private, dedicated API endpoints.",
   "note": "not offered: no provider-side outbound allow/deny-list or no-internet mode documented.",
   "checked": "2026-09-30"
  },
  "egress_http_rules": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Deployments give you production-grade control over your model's infrastructure and provide private, dedicated API endpoints.",
   "note": "not offered.",
   "checked": "2026-09-30"
  },
  "guest_firewall": {
   "value": false,
   "source": "https://cog.run/",
   "quote": "Cog is an open-source tool that lets you package machine learning models in a standard, production-ready container.",
   "note": "derived: isolation is container-based (not a full VM/microVM with guest root), so in-guest firewall control is not applicable.",
   "checked": "2026-09-30"
  },
  "static_egress_ip": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Deployments give you production-grade control over your model's infrastructure and provide private, dedicated API endpoints.",
   "note": "not offered: no fixed outbound IP feature documented.",
   "checked": "2026-09-30"
  },
  "windows": {
   "value": false,
   "source": "https://cog.run/",
   "quote": "Nvidia base images, efficient caching of dependencies, installing specific Python versions",
   "note": "not offered: outside the product's scope; Cog/Replicate infrastructure is Linux/Docker-based GPU hosting.",
   "checked": "2026-09-30"
  },
  "macos": {
   "value": false,
   "source": "https://cog.run/",
   "quote": "Nvidia base images, efficient caching of dependencies, installing specific Python versions",
   "note": "not offered: outside the product's scope; Linux GPU/CPU containers only.",
   "checked": "2026-09-30"
  },
  "arm64": {
   "value": false,
   "source": "https://replicate.com/pricing",
   "quote": "Run your model on Replicate",
   "note": "not offered: published hardware tiers are x86 CPU and NVIDIA GPU (T4/L40S/A100/H100/H200); no ARM CPU option documented.",
   "checked": "2026-09-30"
  },
  "self_host": {
   "value": true,
   "source": "https://cog.run/",
   "quote": "You can deploy your packaged model to your own infrastructure, or to Replicate.",
   "note": "Cog, the packaging/runtime tool underlying Replicate's custom models, is explicitly self-hostable on your own infrastructure.",
   "checked": "2026-09-30"
  },
  "byoc": {
   "value": false,
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Deployments give you production-grade control over your model's infrastructure and provide private, dedicated API endpoints.",
   "note": "not offered: no managed-service-into-your-own-cloud-account option documented; self-hosting Cog yourself is separate from BYOC.",
   "checked": "2026-09-30"
  },
  "open_source": {
   "value": true,
   "source": "https://cog.run/",
   "quote": "Cog is an open-source tool that lets you package machine learning models in a standard, production-ready container.",
   "note": "Cog (github.com/replicate/cog) is open source (Apache-2.0); it underlies custom model deployment on Replicate, though the hosted Replicate platform itself is closed-source.",
   "checked": "2026-09-30"
  },
  "eu_data_residency": {
   "value": false,
   "source": "https://replicate.com/docs/topics/site-policy/subprocessors",
   "quote": "AWS Cloud Hosted Infrastructure and Data Hosting United States",
   "note": "Primary infrastructure/data-hosting subprocessors are listed as United States; no EU-region residency option is documented.",
   "checked": "2026-09-30"
  },
  "max_session_h": {
   "value": "none",
   "source": "https://replicate.com/docs/topics/deployments",
   "quote": "Always-on: Keep instances warm to eliminate cold starts",
   "note": "No instance cap: dedicated/always-on deployment instances run continuously until you scale down or delete them. Each unit of work is capped, though: \"Predictions time out after running for 30 minutes.\" (https://replicate.com/docs/topics/predictions/lifecycle); longer timeouts only by contacting Replicate (\"If you require more than 30 minute timeouts for predictions, contact us.\").",
   "checked": "2026-10-14"
  },
  "sso": {
   "value": false,
   "source": "https://replicate.com/signin",
   "quote": "Sign in with GitHub",
   "note": "The account sign-in page only offers GitHub OAuth login; no SAML/OIDC enterprise SSO option is present there, and no docs page (Organizations, Security/API tokens, Enterprise) mentions SAML, OIDC, or single sign-on. Enterprise page lists priority support, SLAs, higher GPU limits, dedicated account manager, volume discounts, and custom-model help as enterprise perks, but not SSO.",
   "checked": "2026-09-30"
  },
  "cpu_class": {
   "value": {
    "private-cpu": "dedicated",
    "serverless-t4-0": "dedicated",
    "serverless-a100-80gb-1": "dedicated",
    "serverless-h100-2": "dedicated",
    "serverless-l40s-3": "dedicated",
    "serverless-a100-80gb-2x": "dedicated",
    "serverless-l40s-2x": "dedicated",
    "committed-multi-gpu": "dedicated"
   },
   "source": "https://replicate.com/pricing",
   "quote": "Unlike public models, most private models (with the exception of fast booting fine-tunes) run on dedicated hardware so you don't have to share a queue with anyone else. This means you pay for all the time instances of the model are online: the time they spend setting up; the time they spend idle, waiting for requests; and the time they spend active, processing your requests.",
   "note": "Custom-model (Cog) deployments and GPU serverless instances are billed as dedicated per-instance hardware (fixed vCPU/RAM/GPU counts per published hardware tier, e.g. cpu-small=1 vCPU/2GB, gpu-a100-large=10 vCPU/144GB) that is exclusively yours while online (setup+idle+active billed), not a shared/oversubscribed pool. No explicit CPU-sharing or burstable-baseline percentage is published, so this is inferred as dedicated across all listed pricing modes rather than confirmed per-mode with a distinct quote.",
   "checked": "2026-09-30"
  },
  "usage_api": {
   "value": false,
   "source": "https://replicate.com/docs/reference/http",
   "quote": "Predictions return a metrics object with predict_time (seconds), not cost; billing/usage analytics available only via web dashboard, not via REST API",
   "note": "granularity: per-sandbox; cost: no; Replicate's HTTP API provides metrics (predict_time and total_time in seconds) for individual predictions and deployment monitoring (latency, throughput, error rates), but no endpoint returns billable usage or cost data. The account.get endpoint returns account info only. Usage tracking is documented as a web dashboard feature, not a programmatic API.",
   "confidence": "high",
   "checked": "2026-10-01"
  },
  "zero_data_retention": {
   "value": null,
   "source": "https://replicate.com/docs/topics/predictions/data-retention",
   "quote": "all input parameters, output values, output files, and logs are automatically removed after an hour, by default.",
   "note": "Short default retention for API predictions only (web-created predictions are kept indefinitely); no documented zero-retention guarantee or way to change the default, so left unknown rather than true.",
   "checked": "2026-10-14"
  }
 },
 "locations": [],
 "locations_source": "https://replicate.com/pricing",
 "locations_quote": "Replicate does not publish a list of customer-choosable regions or locations for model deployments.",
 "locations_checked": "2026-10-01"
}