{ "features_resolved": { "snapshot_any": true, "snapshot_mem": true, "fork": false, "pause_resume": true, "persistent_disk": true, "whole_disk": null, "persist_scope": null, "volumes": false, "auto_stop_idle": false, "long_sessions": true, "custom_image": true, "image_snapshot": true, "image_oci": true, "vm_isolation": true, "docker_inside": false, "nested_virt": false, "root": null, "systemd": null, "browser": true, "desktop": true, "computer_use": true, "code_interpreter": true, "browser_control": null, "desktop_control": null, "browser_and_desktop": true, "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": false, "egress_allowlist": true, "open_internet": true, "static_egress_ip": false, "private_network": null, "self_host": false, "byoc": false, "open_source": false, "soc2": true, "hipaa": false, "sso": true, "gpu": false, "arm64": false, "windows": false, "macos": false, "lang_python": true, "lang_node": true, "lang_go": true, "lang_rust": null, "lang_java": null, "gpu_desktop": null, "credential_injection": false, "egress_http_rules": false, "wake_on_request": false, "live_resize": false, "fork_running": false, "memory_snapshot_fork": false, "webhooks": true, "scheduled_wakeups": null, "audit_logs": null, "eu_data_residency": true, "zero_data_retention": null, "scoped_api_keys": null, "spend_limits": null, "mcp_server": true, "usage_api": true, "snapshot_auto": false, "snapshot_on_demand": true, "harness_api": false, "own_agent_api": false, "fast_boot": false, "strong_isolation": true, "ingress_rules": false, "guest_firewall": true, "secret_proxy": false, "secret_proxy_any": false, "volume_attach": false, "volume_shared": false, "sidecar": null, "fuse": true, "kernel_features": true, "burst": null }, "company": { "id": "hopx", "checked": "2026-09-30", "company": "Bunnyshell (product: HopX)", "ownership": { "value": "subsidiary", "parent": "Bunnyshell", "source": "https://www.bunnyshell.com", "quote": "Preview, staging, production — and now AI sandboxes. One platform from first commit to production workloads. AI Sandboxes hopx.ai New — Isolated sandboxes for AI agents, scripts, and jobs." }, "stage": { "value": "seed", "source": "https://www.bunnyshell.com/about", "quote": "We are backed by Early Game Ventures, an early-stage venture capital fund focused on enterprise software and deep-tech startups in Central and Eastern Europe.", "confidence": "low", "note": "HopX is a product line of Bunnyshell (branded '[HopX by Bunnyshell]' and listed as 'AI Sandboxes hopx.ai New' on Bunnyshell's own homepage), not a separately-funded entity. No HopX-specific funding was found. This stage reflects Bunnyshell's only disclosed backer, an early-stage VC fund; no specific round type, amount or date is published anywhere found, so this is an inference rather than a confirmed 'seed round'." }, "accelerators": [], "total_funding_usd": { "value": null, "status": "unknown" }, "last_round": { "type": "unknown", "amount_usd": null, "date": null, "leads": [], "status": "unknown", "source": "", "quote": "" }, "investors": [ "Early Game Ventures" ], "portfolio_listings": [], "revenue": { "value_usd": 3600000, "kind": "annual revenue", "as_of": "2026", "status": "estimated", "basis": "Growjo aggregator estimate for Bunnyshell (the parent company); no HopX-specific revenue figure exists, and Growjo revenue estimates are not verified by the company.", "source": "https://growjo.com/company/Bunnyshell", "quote": "Bunnyshell's estimated annual revenue is currently $3.6M per year." }, "employees": { "value": 31, "source": "https://growjo.com/company/Bunnyshell" }, "note": "HopX (hopx.ai) is not an independent company: it is an AI-sandbox product line of Bunnyshell, a Bucharest-founded (2018) DevOps/ephemeral-environments platform (CEO/founder Alin Dobra, also listed as HopX blog author). No YC or other accelerator page was found for 'hopx' or 'bunnyshell' (ycombinator.com/companies/hopx returns 404). Bunnyshell's only disclosed investor is Early Game Ventures (Central/Eastern Europe early-stage VC); no round size, date or valuation is public. Separately, the hopx.ai card's own caveats flag that the HopX Public API was reported down since ~2026-09-27, with no shutdown announced — worth re-checking before relying on this provider. Standard web/DDG/Bing searches in this environment returned unrelated or blocked results, limiting the ability to corroborate funding specifics beyond the primary company sites." }, "id": "hopx", "name": "Hopx", "url": "https://hopx.ai/pricing", "category": "agent-sandbox", "isolation": "firecracker", "modes": [ { "key": "on-demand", "label": "Pay-per-use (per second, allocated, template-fixed size)", "flags": [], "os": [ "linux" ], "regions": [ "us", "eu" ], "gpu_only": false, "features": { "isolation": "firecracker" }, "pricing": "resource", "vcpu_h": 0.0504, "ram_gib_h": 0.0162, "cpu_basis": "alloc", "ram_basis": "alloc", "active_floor": 0, "ram_per_vcpu": [ null, null ], "ram_fixed_per_vcpu": null, "vcpu_options": null, "min_vcpu": 1, "max_vcpu": 64, "max_ram_gib": 62.5, "sizes": null, "pool_tiers": null, "requires_always_on": false, "always_on_month_per_instance": null, "commit_note": null, "min_billed_seconds": 0, "granularity_s": 1, "start_fee": 0, "boot_overhead_s": null, "multiplier": 1, "windows_vcpu_h": null, "gpu": null, "note": "$0.000014/vCPU-s, $0.0000045/GiB-s RAM, $0.00000003/GiB-s disk (identical to E2B). Billed on allocation while running. The page's examples compute on allocated wall-clock, despite the 'pay only when your code runs' marketing. Resources are fixed per template (1-64 vCPU, 512-64,000 MB RAM, 1-250 GB disk; per-plan maxima unpublished). No idle auto-stop: timeout_seconds (default 3,600 s) is a wall-clock auto-delete. Pause stops compute billing. us-east / eu-west have the same price. Basis verified 2026-09-28: Pricing page: \"Pay Only When Your Code Runs\" / \"No idle charges\" (https://hopx.ai/pricing), but its worked examples (e.g. \"Full development day (8 hours active, 2 vCPU, 4GB RAM): ~$1.50\") match allocated vCPU/RAM x wall-clock. Kept alloc." } ], "storage": { "included_disk_gib": 0, "disk_gib_month": 0.0788, "disk_billed_when_stopped": null, "snapshot_gib_month": null, "paused_ram_billed": false }, "network": { "egress_free_gib": null, "egress_gib": null, "ipv4_month": null, "ipv4_available": false, "internet": { "default": "configurable", "evidence": [ { "quote": "\"internet_access\": true", "url": "https://docs.hopx.ai/llms-full.txt" } ] } }, "plans": [ { "name": "Pay-per-use", "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": "The only self-serve offer and the production plan, not a capped trial: hopx.ai/pricing (2026-10-14) shows one \"Pay-Per-Use Pricing\" model with \"Unlimited sandboxes\" and \"No feature gates or tier restrictions\"; the rate-limit docs label it \"Free Tier - Default limits\". No subscription, no minimum. Rate limits: 20 creates/min, 100 control-plane req/min, 300 VM-agent req/min, 10 template builds/h. A sample CLI output shows '10 sandboxes' for Free (unverified). Terms: billed monthly in arrears unless otherwise agreed; the CLI also documents a prepaid balance with optional auto-recharge. Charges non-refundable; fees may change with 30 days' notice. Start rate checked 2026-10-14 (https://docs.hopx.ai/api/concepts/rate-limits): \"| Sandbox Creation | 20 sandboxes | per minute |\" and \"| `/v1/sandboxes` (POST) | 20/minute | Prevent abuse |\", so max_starts_per_min = 20 (a burst of 1,000 takes ~50 min to start).", "max_starts_per_min": 20 }, { "name": "Pro", "fee": null, "flags": [], "fee_is_credit": null, "included_usd": null, "per_seat": null, "concurrency": null, "max_session_h": null, "max_vcpu": null, "max_ram_gib": null, "sizes_allowed": null, "note": "Only in the rate-limit docs ('2x default limits'); not on the pricing page. An illustrative `hopx usage plans` CLI output lists 'Pro $29/mo, 100 sandboxes'. That is unverified, so fee is null. Start rate checked 2026-10-14: \"**Pro Tier** - 2x default limits\" against the default \"Sandbox Creation | 20 sandboxes | per minute\" (https://docs.hopx.ai/api/concepts/rate-limits), so max_starts_per_min = 40 (derived; the page does not print 40).", "max_starts_per_min": 40 }, { "name": "Enterprise", "fee": null, "flags": [ "sales" ], "fee_is_credit": null, "included_usd": null, "per_seat": null, "concurrency": null, "max_session_h": null, "max_vcpu": null, "max_ram_gib": null, "sizes_allowed": null, "note": "Custom SLAs, dedicated support, volume discounts, custom rate limits. Unpublished." } ], "free": { "monthly_credit": 0, "one_time_credit": 200 }, "features": { "linux": true, "windows": false, "macos": false, "arm64": false, "gpu": false, "gpu_types": [], "snapshot": "mem", "pause_resume": true, "fork": false, "persistent_disk": true, "volumes": false, "auto_stop_idle": false, "max_session_h": null, "custom_image": true, "docker_inside": false, "nested_virt": false, "root": null, "systemd": null, "browser": true, "desktop": true, "computer_use": true, "code_interpreter": true, "agent_harnesses": false, "ssh": false, "public_ipv4": false, "inbound_https": true, "custom_domain": null, "raw_tcp_inbound": false, "egress_allowlist": true, "static_egress_ip": false, "private_network": null, "self_host": false, "byoc": false, "open_source": false, "regions": [ "us", "eu" ], "soc2": true, "hipaa": false, "sso": true, "lang_python": true, "lang_node": true, "lang_go": true, "lang_rust": null, "lang_java": null, "gpu_desktop": null, "credential_injection": false, "egress_http_rules": false, "wake_on_request": false, "live_resize": false, "fork_running": false, "memory_snapshot_fork": false, "webhooks": true, "scheduled_wakeups": null, "audit_logs": null, "eu_data_residency": true, "zero_data_retention": null, "scoped_api_keys": null, "spend_limits": null, "mcp_server": true, "isolation": "firecracker", "image_oci": true, "snapshot_auto": false, "snapshot_on_demand": true, "ingress_rules": false, "guest_firewall": true, "secret_proxy": false, "volume_attach": false, "volume_shared": false, "anti_bot": false, "captcha_solving": false, "egress_ip_type": "datacenter", "harness_api": false, "own_agent_api": false, "max_session_unlimited": true, "fast_boot": false, "open_internet": true, "usage_api": true }, "perf": { "cpu_runs_s": null, "cold_start_ms": 4245.82, "burst_tti_ms": 69331, "dax_s": null }, "perf_source": "ComputeSDK (github.com/computesdk/benchmarks results/): cold 4,245.82 ms and burst 69,331 ms, both from 2026-07-17 (stale; low success). No DAX or HPC result.", "knobs": [ { "id": "running_pct", "label": "Share of session running (pause when idle)", "kind": "slider", "min": 0.05, "max": 1, "step": 0.05, "default": 1, "unit": "%", "options": null, "effect": "hours_factor", "modes": [ "on-demand" ], "why": "Hopx bills allocated vCPU/RAM while running, idle included ('no idle charges' means paused); POST /pause stops compute billing (paused storage price unpublished). Pausing between steps bills only the running share.", "source": "https://docs.hopx.ai/api/control-plane/pause-sandbox" }, { "id": "timeout_tail", "label": "Left to the delete timer (x session hours)", "kind": "slider", "min": 1, "max": 4, "step": 0.05, "default": 1, "unit": "x", "options": null, "effect": "hours_factor", "modes": [ "on-demand" ], "why": "There is no idle auto-stop. timeout_seconds is a wall-clock auto-delete from creation. The Python SDK reference says the default is NO timeout (the 3,600 s figure only appears in examples), so a sandbox that is not killed or paused explicitly bills its full allocation indefinitely. The console advertises 'unlimited session time'. Tail factor = max(session, timeout) / session when a timeout is set, and unbounded otherwise.", "source": "https://docs.hopx.ai/core-concepts/sandboxes/timeout" } ], "caveats": [ "CPU billing basis is not stated. The rates are labelled per allocation and the page examples compute on allocated time, so this card treats CPU as alloc. 'Pay only when your code runs' refers to paused/deleted sandboxes, not idle-running ones.", "The pricing-page example '1 vCPU/1 GB/10 GB for 1 h = $0.05' is inconsistent with the published rates, which give $0.0677.", "Paused/stopped storage, snapshot storage and egress prices are unpublished (null). The pricing page claims 'no idle charges' for paused sandboxes.", "No idle auto-stop. The timeout (default 1 h, extendable) deletes the sandbox and its data. Idle-running sandboxes bill full allocation until paused, killed or deleted by the timer.", "Sizes are template-fixed (set at build time). Per-plan vCPU/RAM caps and concurrency caps are unpublished. A CLI sample shows Free = 10 sandboxes, Pro = 100 (unverified).", "Pro plan price is unpublished on the pricing page ($29/mo appears only in illustrative CLI output). Enterprise volume discounts are unpublished.", "The 'gpu' template feature flag exists, but no GPU types or prices are documented.", "Benchmarks: computesdk cold-start success rate is 62% (median 4.2 s), burst success 6% (median 69 s), and DAX never succeeded. Latest runs are 0% successful. No hpc CPU benchmark.", "Included disk is assumed 0. Disk is billed per GiB-second on the template's allocated disk while running.", "unverified: may bill active CPU. hopx.ai/pricing markets \"Pay Only When Your Code Runs\" / \"No idle charges\"; docs and examples indicate allocated resources while running (idle = paused). alloc kept.", "Re-checked 2026-09-28: hopx.ai/pricing lists only vCPU, RAM and storage per-second rates; no snapshot/paused-storage or egress price is published (snapshot_gib_month and egress_gib stay null). Browser: hopx.ai advertises \"Optional browser automation (Firefox/Chromium)\" with VNC/noVNC in the desktop template, so features.browser=true.", "The $200 sign-up credit is one-time and 'may expire under posted program rules' (no expiry date published).", "The hopx.ai site footer shows a static 'All systems operational' badge even while status.hopx.ai shows the Public API down. Rely on status.hopx.ai.", "Activity signals: last product-update blog post is 2025-11-25 (MCP server); GitHub hopx-ai/hopx last push 2026-07-09, cli 2026-05-31. The sandbox heartbeat monitor on the status page is paused.", "Availability warning (2026-09-27 to 2026-10-14, still ongoing): the HopX Public API has been down since about 2026-09-27 07:05 UTC. Re-checked 2026-10-14: status.hopx.ai shows \"Some systems down\", \"HopX Public API | 41.922% Down\", \"Overall Uptime 0.000% Last 24 hours 0.000% Last 7 days 41.922% Last 30 days 80.641% Last 90 days\" and \"There are no updates in the last 30 days.\"; the sandbox monitor is \"Not monitored\" (paused); https://api.hopx.dev/health still returns HTTP 526 (Cloudflare invalid SSL). About 2.5 weeks of outage with no status update, no shutdown notice, GitHub hopx-ai/hopx last pushed 2026-07-09 and the newest blog posts from Nov 2025: treat Hopx as currently unusable. The prices on this card describe the published offer, not a service you can sign up for and run today." ], "sources": [ "https://hopx.ai/pricing", "https://docs.hopx.ai/llms-full.txt", "https://docs.hopx.ai/api/concepts/rate-limits", "https://docs.hopx.ai/cli/commands/billing", "https://docs.hopx.ai/cli/commands/usage", "https://docs.hopx.ai/core-concepts/templates/configuration", "https://docs.hopx.ai/core-concepts/sandboxes/timeout", "https://docs.hopx.ai/core-concepts/sandboxes/managing-state", "https://docs.hopx.ai/api/control-plane/pause-sandbox", "https://docs.hopx.ai/api/control-plane/create-sandbox", "https://www.computesdk.com/benchmarks/sandboxes/" ], "fleet": { "default_max_instances": null, "default_max_vcpu": null, "default_max_gpus": null, "raise": "ticket", "raise_time_days": null, "verification": "Signup advertises $200 credits with no credit card.", "tiers": "Pay-per-use pricing advertises unlimited sandboxes; rate-limit guide mentions higher tiers and support for custom rates, without a numeric account-concurrency ladder.", "notes": "20 creates/minute (Pro 2x = 40) is a launch rate, not a count of running sandboxes; it is now modelled as plans[].max_starts_per_min (2026-10-14). Numeric account fleet quota remains null; published scalability and unlimited wording do not reserve capacity.", "sources": [ "https://hopx.ai/pricing", "https://docs.hopx.ai/api/concepts/rate-limits" ], "researched": "2026-09-28" }, "audit_2026_09_29": { "applied": 3, "source": "checked against the provider's live pricing and docs pages", "reviewed": 3 }, "feature_evidence": { "snapshot_auto": { "value": false, "source": "https://docs.hopx.ai/core-concepts/sandboxes/timeout", "quote": "When a timeout expires, the sandbox is automatically destroyed. All data in the sandbox is lost. Make sure to save important data before the timeout expires.", "note": "Nothing is saved automatically: there is no idle auto-pause, and the wall-clock timeout deletes the sandbox. Pause (memory kept) and stop (filesystem kept) happen only on an explicit API call. SDK Sandbox.connect() auto-resumes a paused sandbox, but pausing itself is manual.", "checked": "2026-09-29" }, "snapshot_on_demand": { "value": true, "source": "https://docs.hopx.ai/core-concepts/sandboxes/managing-state", "quote": "Pause a running sandbox (preserves full state including memory)", "note": "This is only POST /v1/sandboxes/:id/pause plus resume of the SAME sandbox. Memory and processes are included. There are no named snapshots, no fork, and no booting new sandboxes from a checkpoint (templates are built only from Docker images plus build steps). 'hopx system snapshot' is only a metrics snapshot: the CLI source says 'Get metrics snapshot for a sandbox (alias for metrics command)'.", "checked": "2026-09-29" }, "egress_allowlist": { "value": true, "source": "https://docs.hopx.ai/api/control-plane/create-sandbox", "quote": "| `internet_access` | boolean | No | Enable internet access (default: `true`) |", "note": "This is only an on/off no-internet mode, set at create. There are no domain or IP allowlists, so level is null (neither domain nor ip). The create body's full parameter list is template, region, timeout_seconds, internet_access and env_vars.", "checked": "2026-09-29" }, "egress_http_rules": { "value": false, "source": "https://docs.hopx.ai/api/control-plane/create-sandbox", "quote": "| `internet_access` | boolean | No | Enable internet access (default: `true`) |", "note": "The only network control is a boolean. No HTTP-level rules are documented.", "checked": "2026-09-29" }, "ingress_rules": { "value": false, "source": "https://docs.hopx.ai/sdk/python/sandbox", "quote": "HopX automatically exposes all ports. Use this to get the public URL for any port.", "note": "Every port gets a URL of the form https://-..vms.hopx.dev. There is no documented auth token, IP allowlist or port selection for preview URLs. Only the VM-agent API on the sandbox host needs a JWT.", "checked": "2026-09-29" }, "guest_firewall": { "value": true, "source": null, "quote": null, "note": "It is a microVM (Firecracker-like) with its own kernel. The docs show kernel 5.10.186, /sbin/init as PID 1 and commands running as user root. iptables/nftables are never mentioned.", "checked": "2026-09-29", "derived": true }, "secret_proxy": { "value": false, "source": "https://docs.hopx.ai/api/control-plane/create-sandbox", "quote": "| `env_vars` | object | No | Environment variables to set in sandbox |", "note": "Secrets are only env vars, set at create or through the VM-agent env API, so the plaintext is inside the VM. No credential-injection proxy is documented.", "checked": "2026-09-29" }, "volume_attach": { "value": false, "source": "https://docs.hopx.ai/api/control-plane/create-sandbox", "quote": "Resources (vcpu, memory, disk) are **always loaded from the template** - you cannot specify custom resources.", "note": "The create API has no volume or mount parameter, and the docs never mention volumes. Storage is only the template root disk (1-250 GB).", "checked": "2026-09-29" }, "volume_shared": { "value": false, "source": "https://docs.hopx.ai/api/control-plane/create-sandbox", "quote": "Resources (vcpu, memory, disk) are **always loaded from the template** - you cannot specify custom resources.", "note": "No shared volumes, network FS or bucket mounts are documented.", "checked": "2026-09-29" }, "docker_inside": { "value": false, "source": "https://docs.hopx.ai/llms-full.txt", "quote": "Build custom templates from Docker images and build steps", "note": "The only Docker-related capability documented is building a sandbox TEMPLATE from a Docker image at build time. The extensive VM Agent API docs (code execution, shell commands, process management, filesystem) never mention a Docker/container runtime, dockerd, or docker CLI available inside a running sandbox.", "checked": "2026-09-30" }, "nested_virt": { "value": false, "source": "https://hopx.ai/security", "quote": "Compute isolation: Each sandbox runs inside a Firecracker micro-VM for kernel-level isolation (stronger than containers).", "note": "derived: Firecracker microVMs do not expose /dev/kvm or nested virtualization to the guest by design, and no docs page mentions KVM, nested virtualization, or running VMs/emulators inside a sandbox.", "checked": "2026-09-30" }, "anti_bot": { "value": false, "source": "https://hopx.ai", "quote": "Hopx - Secure Sandboxes for AI Agents | Run Untrusted Code Safely", "note": "not offered: outside the product's scope. Hopx is a code-execution/desktop-automation sandbox (microVMs for running code and optional VNC desktop automation); no stealth, fingerprint-evasion, or anti-bot-detection feature is documented anywhere in the docs.", "checked": "2026-09-30" }, "captcha_solving": { "value": false, "source": "https://hopx.ai", "quote": "Hopx - Secure Sandboxes for AI Agents | Run Untrusted Code Safely", "note": "not offered: outside the product's scope. No CAPTCHA-solving feature, API, or partner integration is mentioned anywhere in the docs or pricing/feature pages.", "checked": "2026-09-30" }, "egress_ip_type": { "value": "datacenter", "source": "https://docs.hopx.ai/api/control-plane/create-sandbox", "quote": "| `internet_access` | boolean | No | Enable internet access (default: `true`) |", "note": "derived: outbound networking is only a boolean on/off toggle; no static, dedicated, residential, or mobile egress IP option is documented, so outbound traffic uses the provider's shared cloud-datacenter IP ranges.", "checked": "2026-09-30" }, "wake_on_request": { "value": false, "source": "https://docs.hopx.ai/llms-full.txt", "quote": "You cannot connect to a stopped sandbox. Use `sandbox.start()` first, or create a new sandbox.", "note": "There is no auto-stop-on-idle in the first place (per the card), and resuming a paused sandbox requires an explicit SDK/API call (Sandbox.connect() or sandbox.start()) rather than an inbound HTTP request to an exposed port automatically waking it.", "checked": "2026-09-30" }, "live_resize": { "value": false, "source": "https://docs.hopx.ai/core-concepts/templates/configuration", "quote": "Resources (vcpu, memory, disk) are **always loaded from the template** - you cannot specify custom resources.", "note": "CPU/RAM are fixed per template at build time and cannot be changed for a running sandbox; no resize/scale API is documented (only terminal and desktop-window resize exist, which are unrelated).", "checked": "2026-09-30" }, "harness_api": { "value": false, "source": "https://hopx.ai/pricing", "quote": "Pre-built Templates Ready-to-use environments for instant start", "note": "not offered: outside the product's scope. Hopx offers generic code-interpreter/desktop templates, not coding-agent harnesses (Claude Code, Codex, OpenCode, etc.) or an API to drive one; none are mentioned anywhere in the docs (matches features.agent_harnesses=false on the card).", "checked": "2026-09-30" }, "own_agent_api": { "value": false, "source": "https://hopx.ai/pricing", "quote": "Automation Desktop Automation Control cloud desktops programmatically", "note": "not offered: outside the product's scope. Hopx exposes a desktop-automation API (VNC, mouse/keyboard, screenshots) for the CUSTOMER's own agent code to drive, but ships no first-party hosted browser/computer-use/research agent of its own.", "checked": "2026-09-30" }, "static_egress_ip": { "value": false, "source": "https://docs.hopx.ai/api/control-plane/create-sandbox", "quote": "| `internet_access` | boolean | No | Enable internet access (default: `true`) |", "note": "The only outbound network control documented is a boolean internet_access toggle; no static/fixed/dedicated outbound IP feature is mentioned anywhere in the docs.", "checked": "2026-09-30" }, "soc2": { "value": true, "source": "https://hopx.ai/security", "quote": "SOC 2 — independent assessment of security, availability, and confidentiality controls over an evaluation period.", "note": "Listed under 'Security at Hopx SOC 2 • ISO/IEC 27001 • ISO 9001:2015'; certificate/bridge letter copies offered under NDA.", "checked": "2026-09-30" }, "hipaa": { "value": false, "source": "https://hopx.ai/security", "quote": "Compliance & Certifications SOC 2 — ... ISO/IEC 27001 — ... ISO 9001:2015 — ...", "note": "not offered: the security/compliance page enumerates SOC 2, ISO/IEC 27001 and ISO 9001:2015 as its certifications and never mentions HIPAA or a BAA anywhere on the page.", "checked": "2026-09-30" }, "sso": { "value": true, "source": "https://hopx.ai/security", "quote": "Customer auth: Support for SSO (e.g., OIDC/SAML) and enforced MFA for privileged actions (where configured).", "note": null, "checked": "2026-09-30" }, "eu_data_residency": { "value": true, "source": "https://hopx.ai/security", "quote": "Residency: Execute and store data in the region(s) you select where supported.", "note": "The card's regions include 'eu' (eu-west), so customers can select an EU region for execution and data residency.", "checked": "2026-09-30" }, "max_session_h": { "value": "none", "source": "https://docs.hopx.ai/llms-full.txt", "quote": "No maximum - There's no maximum timeout limit (subject to your plan)", "note": "The timeout is a wall-clock auto-delete timer, extendable indefinitely via PUT /timeout (docs show a 30-day example under 'Keep Sandbox Alive Indefinitely'), and there is no stated hard cap, so a sandbox can in principle run until the owner stops it.", "checked": "2026-09-30" }, "fork": { "value": false, "source": "https://docs.hopx.ai/core-concepts/sandboxes/managing-state", "quote": "Pause a running sandbox (preserves full state including memory)", "note": "State management is only pause/resume/stop/kill of the SAME sandbox. There is no documented API or CLI verb to clone a sandbox into a new one (no 'fork'/'clone' endpoint anywhere in llms-full.txt or the control-plane API reference).", "checked": "2026-09-30" }, "fork_running": { "value": false, "source": "https://docs.hopx.ai/llms-full.txt", "quote": "(no occurrence of \"fork\" anywhere in the full documentation export)", "note": "Since forking at all is not offered (see fork), forking a running sandbox is a fortiori not offered.", "checked": "2026-09-30" }, "memory_snapshot_fork": { "value": false, "source": "https://docs.hopx.ai/core-concepts/sandboxes/managing-state", "quote": "Pause a running sandbox (preserves full state including memory)", "note": "The only memory-preserving operation is pause/resume of the same sandbox id, not a clone that boots as a new, independent running sandbox. No fork/checkpoint-to-new-sandbox flow is documented.", "checked": "2026-09-30" }, "ssh": { "value": false, "source": "https://docs.hopx.ai/core-concepts/terminal/websocket", "quote": "The WebSocket terminal provides interactive, real-time access to a PTY (pseudo-terminal) in your sandbox.", "note": "Interactive shell access is exposed only via a WebSocket-based PTY terminal (SDK `sandbox.terminal.connect()` / CLI `hopx terminal connect`), reached through the control-plane API with a bearer/JWT token. The string \"ssh\" does not appear anywhere in the full documentation export (docs.hopx.ai/llms-full.txt) or the create-sandbox API reference; there is no SSH key upload or sshd port documented.", "checked": "2026-09-30" }, "auto_stop_idle": { "value": false, "source": "https://docs.hopx.ai/core-concepts/sandboxes/timeout", "quote": "When a timeout expires, the sandbox is automatically destroyed. All data in the sandbox is lost.", "note": "The only automatic lifecycle action is `timeout_seconds`, a fixed wall-clock timer counted from creation (extendable via PUT /timeout) that DELETES the sandbox, not an idle-detecting stop/sleep. A busy or idle sandbox is billed identically until this timer fires or the user explicitly pauses/stops it; no idle-activity detection is documented anywhere in the docs.", "checked": "2026-09-30" }, "volumes": { "value": false, "source": "https://docs.hopx.ai/api/control-plane/create-sandbox", "quote": "Resources (vcpu, memory, disk) are **always loaded from the template** - you cannot specify custom resources.", "note": "The create-sandbox API has no volume/mount parameter, and the string \"volume\" never appears in the docs except in unrelated phrases like \"high-volume\". Storage is only the single template-defined root disk (1-250 GB); there is no separate attachable or shared volume primitive.", "checked": "2026-09-30" }, "public_ipv4": { "value": false, "source": "https://docs.hopx.ai/sdk/python/sandbox", "quote": "HopX automatically exposes all ports. Use this to get the public URL for any port.", "note": "Network access to a sandbox's ports is only via generated HTTPS subdomain URLs of the form https://-..vms.hopx.dev; the term \"public IP\"/\"ipv4\"/\"static ip\" never appears in the documentation export, and the pricing/network fields for ipv4 are already unpublished/false on the card.", "checked": "2026-09-30" }, "agent_harnesses": { "value": false, "source": "https://docs.hopx.ai/core-concepts/templates/listing", "quote": "Templates include both public templates (available to all users) and organization-specific templates (custom templates you've built). ... Filter by Category ... category=\"development\" ... category=\"infrastructure\" ... category=\"operating-system\"", "note": "Template categories are development/infrastructure/operating-system and per-language (python, nodejs, etc.); none are coding-agent harness templates. The docs export never mentions Claude Code, Codex, OpenCode, or any preinstalled coding-agent harness (the only 'Codex' hit is a generic IDE/MCP-client name in an unrelated integrations list).", "checked": "2026-09-30" }, "windows": { "value": false, "source": "https://docs.hopx.ai/api/vm-agent/get-system-metrics", "quote": "\"os\": {\"name\": \"Ubuntu 22.04.5 LTS\", \"version\": \"22.04.5 LTS (Jammy Jellyfish)\", \"kernel\": \"5.10.186\", \"arch\": \"amd64\"}", "note": "Sandboxes run Ubuntu Linux inside Firecracker microVMs. Every occurrence of 'Windows' in the docs refers to the local client OS (CLI install instructions, IDE keyboard shortcuts, MCP client config paths for Claude Desktop) not the sandbox guest OS. Windows guests are outside the product's scope.", "checked": "2026-09-30" }, "macos": { "value": false, "source": "https://docs.hopx.ai/api/vm-agent/get-system-metrics", "quote": "\"os\": {\"name\": \"Ubuntu 22.04.5 LTS\", ...}", "note": "Same as windows: sandboxes are Ubuntu Linux Firecracker microVMs. Every 'macOS' mention in the docs is about the local developer machine (CLI/SDK install, keyboard shortcuts). macOS guests are outside the product's scope.", "checked": "2026-09-30" }, "arm64": { "value": false, "source": "https://docs.hopx.ai/api/vm-agent/get-system-metrics", "quote": "\"kernel\": \"5.10.186\", \"arch\": \"amd64\"", "note": "The documented sandbox system-info response reports arch=amd64. No arm64/aarch64 sandbox architecture option is documented anywhere in the docs export.", "checked": "2026-09-30" }, "self_host": { "value": false, "source": "https://docs.hopx.ai/llms-full.txt", "quote": "(no occurrence of \"self-host\", \"on-prem\", \"on-premises\", or \"own infrastructure\" anywhere in the full documentation export)", "note": "HopX is offered only as HopX's own hosted multi-tenant cloud service (api.hopx.dev / vms.hopx.dev); no self-hosted/on-prem deployment option is documented.", "checked": "2026-09-30" }, "byoc": { "value": false, "source": "https://docs.hopx.ai/llms-full.txt", "quote": "(no occurrence of \"BYOC\", \"bring your own cloud\", \"your own account\", or \"VPC peering\" anywhere in the full documentation export)", "note": "Only two regions (us-east, eu-west) on HopX's own infrastructure are offered at sandbox create; there is no documented option to deploy the control plane into a customer's own cloud account.", "checked": "2026-09-30" }, "open_source": { "value": false, "source": "https://github.com/hopx-ai/hopx", "quote": "The Cloud for AI Agents — secure, real-world environments that spin up in milliseconds. (repo contents: cli/, cookbook/, javascript/, python/, openapi_agent.json, openapi_public.json)", "note": "The hopx-ai/hopx GitHub repo (MIT license) contains only client SDKs (Python, JavaScript), the CLI, cookbooks and OpenAPI specs — not the core sandbox/VM orchestration engine (the Firecracker control plane and VM-agent), which is not published. Client SDKs being MIT-licensed does not make the core product open source.", "checked": "2026-09-30" }, "gpu": { "value": false, "source": "https://docs.hopx.ai/llms-full.txt", "quote": "* `gpu` - GPU support (for ML/AI workloads)", "note": "The docs list `gpu` as one of a template's boolean 'features' flags, but this is the only mention of GPU anywhere in the full docs export, pricing page, or security page: no GPU instance types, GPU pricing, GPU regions, or any template that actually sets this flag are published anywhere. The card's own on-demand mode has gpu=null/gpu_only=false and the caveats note 'the gpu template feature flag exists, but no GPU types or prices are documented.' Treated as not actually offered as a purchasable product today.", "checked": "2026-09-30" }, "fast_boot": { "value": false, "source": "https://www.computesdk.com/benchmarks", "quote": "burst of 100: p95 70411 ms, 6% started", "note": "ComputeSDK sandbox benchmark 2026-07-17", "checked": "2026-07-17" }, "usage_api": { "value": true, "source": "https://docs.hopx.ai/cli/commands/usage", "quote": "hopx usage sandboxes: View sandbox usage statistics. Expected Output shows Sandbox ID, Runtime, Resources per individual sandbox", "note": "granularity: per-sandbox; cost: no; CLI command provides per-sandbox runtime and resource usage; billing commands provide account-level cost data", "confidence": "high", "checked": "2026-10-01" } }, "locations": [], "locations_source": "https://hopx.ai/pricing and https://docs.hopx.ai/api/control-plane/create-sandbox", "locations_quote": "Region parameter documented in API; caveats note 'us-east / eu-west have the same price'. Specific country or city locations not documented in public docs or pricing pages.", "locations_checked": "2026-10-01" }