{
"category": "agent-sandboxes",
"rows": [
{
"basis": "Running time",
"caveats": [],
"cost": "≥ $0.017",
"cost_usd": 0.017,
"exact_shape": true,
"fleet": "≥ $149.60",
"fleet_plan": null,
"fleet_usd": 149.60000000000002,
"light": "≥ $0.017",
"light_lower": false,
"lower_bound": true,
"notes": [
"minimum"
],
"per_hour": "$0.017",
"provider": {
"bird": "",
"bowerbird": false,
"categories": [
{
"href": "/agent-sandboxes/",
"label": "Agent sandboxes",
"slug": "agent-sandboxes"
},
{
"href": "/devboxes/",
"label": "Devboxes",
"slug": "devboxes"
}
],
"href": "/providers/instabox/",
"logo": "/logos/instabox.svg",
"logo_dark": false,
"name": "InstaBox",
"operator": "InstaBox",
"plate": 26,
"slug": "instabox",
"status": "active",
"status_label": "Active",
"status_since": null,
"summary": "Firecracker microVM machines for agents with a desktop, Docker and coding agents preinstalled, billed per second for vCPU and memory on top of a monthly plan."
},
"rank": 1,
"shape": "2 vCPU / 4 GB",
"src": 1,
"terms": {
"at_once": {
"text": "100"
},
"pay": {
"detail": "A monthly plan from $20/mo that includes usage; more usage costs extra.",
"key": "subscription",
"label": "Plan from $20/mo"
},
"starts": {
"detail": "Sandbox starts (create, resume, fork) allowed: 10/min, then refused.",
"over": "then refused",
"text": "10/min"
}
},
"vt": true
},
{
"basis": "Running time",
"caveats": [],
"cost": "≥ $0.018",
"cost_usd": 0.018,
"exact_shape": true,
"fleet": "≥ $158.40",
"fleet_plan": null,
"fleet_usd": 158.39999999999998,
"light": "≥ $0.018",
"light_lower": false,
"lower_bound": true,
"notes": [
"minimum"
],
"per_hour": "$0.018",
"provider": {
"bird": "",
"bowerbird": false,
"categories": [
{
"href": "/ci-runners/",
"label": "CI runners",
"slug": "ci-runners"
},
{
"href": "/agent-sandboxes/",
"label": "Agent sandboxes",
"slug": "agent-sandboxes"
},
{
"href": "/devboxes/",
"label": "Devboxes",
"slug": "devboxes"
}
],
"href": "/providers/boat/",
"logo": "/logos/boat.svg",
"logo_dark": false,
"name": "boat",
"operator": "boat",
"plate": 6,
"slug": "boat",
"status": "active",
"status_label": "Active",
"status_since": null,
"summary": "Full-VM sandboxes for coding agents on a monthly plan, plus hosted GitHub Actions runners on the same machines."
},
"rank": 2,
"shape": "small (2 vCPU / 4 GB)",
"src": 2,
"terms": {
"at_once": {
"text": "100"
},
"pay": {
"detail": "A monthly plan from $20/mo that includes usage; more usage costs extra.",
"key": "subscription",
"label": "Plan from $20/mo"
},
"starts": {
"detail": "Sandbox starts (create, resume, fork) allowed: 12/min · 60/h · 200/day, then from $0.03 each.",
"over": "then from $0.03 each",
"text": "12/min · 60/h · 200/day"
}
},
"vt": true
},
{
"basis": "Running time",
"caveats": [],
"cost": "≥ $0.0667",
"cost_usd": 0.06666,
"exact_shape": true,
"fleet": "≥ $586.61",
"fleet_plan": null,
"fleet_usd": 586.608,
"light": "≥ $0.0667",
"light_lower": false,
"lower_bound": true,
"notes": [
"minimum"
],
"per_hour": "$0.0667",
"provider": {
"bird": "",
"bowerbird": false,
"categories": [
{
"href": "/agent-sandboxes/",
"label": "Agent sandboxes",
"slug": "agent-sandboxes"
}
],
"href": "/providers/northflank/",
"logo": "/logos/northflank.png",
"logo_dark": false,
"name": "Northflank",
"operator": "Northflank Ltd",
"plate": 35,
"slug": "northflank",
"status": "active",
"status_label": "Active",
"status_since": null,
"summary": "Sandboxes on Northflank's container platform, in microVMs for CPU workloads and gVisor for GPUs, billed per second for vCPU and memory on Northflank Cloud or in your own cloud."
},
"rank": 3,
"shape": "2 vCPU / 4 GB",
"src": 3,
"terms": {
"at_once": {
"text": "500"
},
"pay": {
"detail": "Billed for what you use, with no plan fee.",
"key": "payg",
"label": "Pay as you go"
},
"starts": {
"detail": "No limit on starting sandboxes is published.",
"over": "",
"text": "?"
}
},
"vt": true
},
{
"basis": "Active CPU",
"caveats": [
{
"detail": "Only CPU in use is billed, so idle time costs less; the light-use column shows how much.",
"key": "active-cpu",
"label": "Active-CPU billing"
}
],
"cost": "≥ $0.1",
"cost_usd": 0.1,
"exact_shape": true,
"fleet": "≥ $880.00",
"fleet_plan": null,
"fleet_usd": 880.0,
"light": "≥ $0.01",
"light_lower": true,
"lower_bound": true,
"notes": [
"rounding",
"minimum"
],
"per_hour": "$0.1",
"provider": {
"bird": "",
"bowerbird": false,
"categories": [
{
"href": "/agent-sandboxes/",
"label": "Agent sandboxes",
"slug": "agent-sandboxes"
},
{
"href": "/devboxes/",
"label": "Devboxes",
"slug": "devboxes"
}
],
"href": "/providers/upstash-box/",
"logo": "/logos/upstash-box.png",
"logo_dark": false,
"name": "Upstash Box",
"operator": "Upstash, Inc.",
"plate": 45,
"slug": "upstash-box",
"status": "active",
"status_label": "Active",
"status_since": null,
"summary": "Container sandboxes with Claude Code, Codex, OpenCode or Cursor built in, billed per active CPU hour by size, with fixed monthly keep-alive boxes."
},
"rank": 4,
"shape": "Small (2 vCPU / 4 GB)",
"src": 4,
"terms": {
"at_once": {
"text": "1,000"
},
"pay": {
"detail": "Billed for what you use, with no plan fee.",
"key": "payg",
"label": "Pay as you go"
},
"starts": {
"detail": "No limit on starting sandboxes is published.",
"over": "",
"text": "?"
}
},
"vt": true
},
{
"basis": "Running time",
"caveats": [],
"cost": "≥ $0.117",
"cost_usd": 0.11664,
"exact_shape": true,
"fleet": "≥ $1,026",
"fleet_plan": null,
"fleet_usd": 1026.432,
"light": "≥ $0.117",
"light_lower": false,
"lower_bound": true,
"notes": [
"minimum"
],
"per_hour": "$0.117",
"provider": {
"bird": "",
"bowerbird": false,
"categories": [
{
"href": "/agent-sandboxes/",
"label": "Agent sandboxes",
"slug": "agent-sandboxes"
}
],
"href": "/providers/novita/",
"logo": "/logos/novita.png",
"logo_dark": false,
"name": "Novita AI",
"operator": "Novita AI",
"plate": 36,
"slug": "novita",
"status": "active",
"status_label": "Active",
"status_since": null,
"summary": "Agent sandboxes with pause/resume, memory snapshots and VNC desktops, billed per second per allocated vCPU and GiB of memory."
},
"rank": 5,
"shape": "2 vCPU / 4 GB",
"src": 5,
"terms": {
"at_once": {
"text": "100"
},
"pay": {
"detail": "Billed for what you use, with no plan fee.",
"key": "payg",
"label": "Pay as you go"
},
"starts": {
"detail": "No limit on starting sandboxes is published.",
"over": "",
"text": "?"
}
},
"vt": true
},
{
"basis": "Active CPU",
"caveats": [
{
"detail": "Only CPU in use is billed, so idle time costs less; the light-use column shows how much.",
"key": "active-cpu",
"label": "Active-CPU billing"
}
],
"cost": "≥ $0.2",
"cost_usd": 0.2,
"exact_shape": true,
"fleet": "≥ $1,760",
"fleet_plan": null,
"fleet_usd": 1760.0,
"light": "≥ $0.074",
"light_lower": true,
"lower_bound": true,
"notes": [
"minimum"
],
"per_hour": "$0.2",
"provider": {
"bird": "",
"bowerbird": false,
"categories": [
{
"href": "/agent-sandboxes/",
"label": "Agent sandboxes",
"slug": "agent-sandboxes"
},
{
"href": "/devboxes/",
"label": "Devboxes",
"slug": "devboxes"
}
],
"href": "/providers/tensorlake/",
"logo": "/logos/tensorlake.svg",
"logo_dark": false,
"name": "Tensorlake",
"operator": "Tensorlake Inc.",
"plate": 42,
"slug": "tensorlake",
"status": "active",
"status_label": "Active",
"status_since": null,
"summary": "MicroVM sandboxes for AI agents with suspend/resume, memory snapshots and clones, billed for active CPU plus allocated memory and disk from prepaid credit or a Pro plan."
},
"rank": 6,
"shape": "2 vCPU / 4 GB",
"src": 6,
"terms": {
"at_once": {
"text": "100"
},
"pay": {
"detail": "You buy credit up front, then use it. Whether unused credit expires is not published.",
"key": "prepaid",
"label": "Prepaid credit"
},
"starts": {
"detail": "No limit on starting sandboxes is published.",
"over": "",
"text": "?"
}
},
"vt": true
},
{
"basis": "Running time",
"caveats": [],
"cost": "≥ $0.206",
"cost_usd": 0.20600000000000002,
"exact_shape": true,
"fleet": "≥ $1,813",
"fleet_plan": null,
"fleet_usd": 1812.8000000000002,
"light": "≥ $0.206",
"light_lower": false,
"lower_bound": true,
"notes": [
"minimum"
],
"per_hour": "$0.206",
"provider": {
"bird": "",
"bowerbird": false,
"categories": [
{
"href": "/agent-sandboxes/",
"label": "Agent sandboxes",
"slug": "agent-sandboxes"
}
],
"href": "/providers/google-agent-sandbox/",
"logo": "/logos/google-agent-sandbox.png",
"logo_dark": false,
"name": "Google Cloud Agent Platform sandboxes",
"operator": "Google LLC",
"plate": 25,
"slug": "google-agent-sandbox",
"status": "active",
"status_label": "Active",
"status_since": null,
"summary": "Managed sandboxes in Google Cloud's Gemini Enterprise Agent Platform (formerly Vertex AI Agent Engine) for code execution, shell, custom containers and computer use, billed per vCPU-hour and GiB-hour."
},
"rank": 7,
"shape": "2 vCPU / 4 GB",
"src": 7,
"terms": {
"at_once": {
"text": "1,000"
},
"pay": {
"detail": "Billed for what you use, with no plan fee.",
"key": "payg",
"label": "Pay as you go"
},
"starts": {
"detail": "No limit on starting sandboxes is published.",
"over": "",
"text": "?"
}
},
"vt": true
},
{
"basis": "Active CPU",
"caveats": [
{
"detail": "Only CPU in use is billed, so idle time costs less; the light-use column shows how much.",
"key": "active-cpu",
"label": "Active-CPU billing"
},
{
"detail": "A sandbox is stopped after 8 h.",
"key": "session-cap",
"label": "Max session 8 h"
}
],
"cost": "$0.217",
"cost_usd": 0.2168,
"exact_shape": true,
"fleet": "$1,908",
"fleet_plan": null,
"fleet_usd": 1907.84,
"light": "$0.0557",
"light_lower": true,
"lower_bound": false,
"notes": [],
"per_hour": "$0.217",
"provider": {
"bird": "",
"bowerbird": false,
"categories": [
{
"href": "/agent-sandboxes/",
"label": "Agent sandboxes",
"slug": "agent-sandboxes"
}
],
"href": "/providers/aws-agentcore/",
"logo": "/logos/aws-agentcore.png",
"logo_dark": false,
"name": "AWS Bedrock AgentCore Code Interpreter",
"operator": "Amazon Web Services, Inc.",
"plate": 2,
"slug": "aws-agentcore",
"status": "active",
"status_label": "Active",
"status_since": null,
"summary": "Managed code-execution sessions for agents in Amazon Bedrock AgentCore, each in its own microVM, billed per second for CPU actually used and peak memory."
},
"rank": 8,
"shape": "2 vCPU / 4 GB",
"src": 8,
"terms": {
"at_once": {
"text": "1,000"
},
"pay": {
"detail": "Billed for what you use, with no plan fee.",
"key": "payg",
"label": "Pay as you go"
},
"starts": {
"detail": "Sandbox starts (create, resume, fork) allowed: 1,800/min · 108,000/h · 2,592,000/day, then refused.",
"over": "then refused",
"text": "1,800/min · 108,000/h · 2,592,000/day"
}
},
"vt": true
},
{
"basis": "Running time",
"caveats": [],
"cost": "≥ $0.238",
"cost_usd": 0.23795999999999998,
"exact_shape": true,
"fleet": "≥ $2,094",
"fleet_plan": null,
"fleet_usd": 2094.048,
"light": "≥ $0.238",
"light_lower": false,
"lower_bound": true,
"notes": [
"minimum"
],
"per_hour": "$0.238",
"provider": {
"bird": "",
"bowerbird": false,
"categories": [
{
"href": "/agent-sandboxes/",
"label": "Agent sandboxes",
"slug": "agent-sandboxes"
}
],
"href": "/providers/modal/",
"logo": "/logos/modal.svg",
"logo_dark": false,
"name": "Modal",
"operator": "Modal Labs, Inc.",
"plate": 32,
"slug": "modal",
"status": "active",
"status_label": "Active",
"status_since": null,
"summary": "Serverless compute platform whose Sandboxes run untrusted or agent-written code in gVisor or VM containers, billed per second for the larger of requested and used CPU and memory."
},
"rank": 9,
"shape": "2 vCPU / 4 GB",
"src": 9,
"terms": {
"at_once": {
"text": "100"
},
"pay": {
"detail": "Billed for what you use, with no plan fee.",
"key": "payg",
"label": "Pay as you go"
},
"starts": {
"detail": "No limit on starting sandboxes is published.",
"over": "",
"text": "?"
}
},
"vt": true
},
{
"basis": "Active CPU",
"caveats": [
{
"detail": "Only CPU in use is billed, so idle time costs less; the light-use column shows how much.",
"key": "active-cpu",
"label": "Active-CPU billing"
},
{
"detail": "A fee of $0.0000006 per sandbox started comes on top; it's in the costs shown.",
"key": "job-fee",
"label": "+$0.0000006 per sandbox"
}
],
"cost": "≥ $0.341",
"cost_usd": 0.3408006,
"exact_shape": true,
"fleet": "≥ $2,999",
"fleet_plan": null,
"fleet_usd": 2999.0406599999997,
"light": "≥ $0.11",
"light_lower": true,
"lower_bound": true,
"notes": [
"rounding"
],
"per_hour": "$0.341",
"provider": {
"bird": "",
"bowerbird": false,
"categories": [
{
"href": "/agent-sandboxes/",
"label": "Agent sandboxes",
"slug": "agent-sandboxes"
}
],
"href": "/providers/vercel-sandbox/",
"logo": "/logos/vercel-sandbox.png",
"logo_dark": false,
"name": "Vercel Sandbox",
"operator": "Vercel Inc.",
"plate": 46,
"slug": "vercel-sandbox",
"status": "active",
"status_label": "Active",
"status_since": null,
"summary": "Firecracker microVM sandboxes started through Vercel's SDKs and CLI, billed for active CPU time plus provisioned memory."
},
"rank": 10,
"shape": "2 vCPU / 4 GB",
"src": 10,
"terms": {
"at_once": {
"text": "10,000"
},
"pay": {
"detail": "Billed for what you use, on a plan of at least $20/mo first.",
"key": "payg",
"label": "Pay as you go, $20/mo min"
},
"starts": {
"detail": "No limit on starting sandboxes is published.",
"over": "",
"text": "?"
}
},
"vt": true
}
],
"scenarios": {
"fleet_days_per_month": 22.0,
"fleet_hours_per_day": 8.0,
"fleet_sandboxes": 50,
"light_cpu_share": 0.1
},
"session_hours": 1.0,
"shape": {
"memory_gb": 4.0,
"vcpu": 2
},
"unranked": []
}