{ "features_resolved": { "snapshot_any": true, "snapshot_mem": true, "fork": true, "pause_resume": true, "persistent_disk": true, "whole_disk": false, "persist_scope": true, "volumes": false, "auto_stop_idle": true, "long_sessions": true, "custom_image": true, "image_snapshot": true, "image_oci": true, "vm_isolation": true, "docker_inside": true, "nested_virt": null, "root": null, "systemd": null, "browser": false, "desktop": true, "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": true, "ssh": true, "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": false, "byoc": false, "open_source": false, "soc2": false, "hipaa": false, "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": false, "egress_http_rules": false, "wake_on_request": false, "live_resize": false, "fork_running": true, "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": false, "usage_api": false, "snapshot_auto": true, "snapshot_on_demand": true, "harness_api": true, "own_agent_api": false, "fast_boot": null, "strong_isolation": true, "ingress_rules": true, "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": "tama", "checked": "2026-09-30", "company": "tama", "ownership": { "value": "independent", "parent": null, "source": "https://tama.computer/", "quote": "Give your agent a machine to hack on and pay by the second.", "note_confidence": "low — no dedicated about/team page was reachable (redirects to homepage); no evidence found of acquisition or parent-company ownership, treated as an independent startup by default." }, "stage": { "value": "unknown", "source": "", "quote": "", "confidence": "low", "note": "No funding announcement, press coverage, Crunchbase/PitchBook profile, or accelerator listing could be found. The product's only public trace is a low-traction Show HN launch (\"Show HN: Agent Sandboxes with GPUs from $0.20/hr\", 5 points, posted by HN user 'Mernit' on 2026-09-04), which does not establish a funding stage." }, "accelerators": [], "total_funding_usd": { "value": null, "status": "unknown" }, "last_round": { "type": null, "amount_usd": null, "date": null, "leads": [], "status": "unknown", "source": "", "quote": "" }, "investors": [], "portfolio_listings": [], "revenue": { "value": null, "status": "unknown" }, "employees": { "value": null, "source": "" }, "note": "tama.computer is a Linux-VM 'computers for agents' sandbox/GPU rental product (CLI + Python/TypeScript SDKs) launched publicly via Show HN on 2026-09-04 under the HN handle 'Mernit'. No about/team/careers/press content is reachable on the site (all redirect to the homepage), and no funding, accelerator, investor, revenue, or employee-count information could be found via Crunchbase (blocked by Cloudflare), YCombinator's company directory (no listing found for 'tama'), or web search (DuckDuckGo, Bing, Google and Mojeek all returned bot-challenge/blocked responses throughout this session, preventing further corroboration within the ~15 minute research budget). Ownership, stage, funding, and revenue are therefore marked unknown/independent-by-default rather than guessed; this should be re-verified when search access is available." }, "id": "tama", "name": "tama", "url": "https://tama.computer/", "category": "agent-sandbox", "isolation": "vm", "modes": [ { "key": "cpu-on-demand", "label": "CPU on-demand", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": false, "features": { "isolation": "vm" }, "pricing": "resource", "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": "Public calculator CPU+RAM rates; allocation, not CPU utilization.", "vcpu_h": 0.01166, "ram_gib_h": 0.005523, "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": 32, "max_ram_gib": 128, "perf": { "cpu_runs_s": 16.77 }, "perf_source": "StarSling hpc-sandbox-benchmarks, run 36498291449 (2026-09-29), Node.js V8 Web Tooling p50 at a 4 vCPU / 8 GB target" }, { "key": "gpu-rtx-4090", "label": "RTX 4090 live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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-4090": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "RTX 4090", "vcpu": 6, "ram_gib": 17, "disk_gib": 50, "hour": 0.54, "month_cap": null } ] }, { "key": "gpu-a-6000", "label": "A 6000 live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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-A6000": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension. 2026-10-14 verify: offers API \"gpu\":\"A6000\",\"hourly_micros\":769500, node 6000 millicores / 49152 MB; homepage \"A6000 48 GB 6 48 GB 1 ... $0.77 /hr\" (was $0.675 on 28 vCPU / 58 GiB).", "sizes": [ { "name": "A 6000", "vcpu": 6, "ram_gib": 48, "disk_gib": 50, "hour": 0.7695, "month_cap": null } ] }, { "key": "gpu-rtx-5090", "label": "RTX 5090 live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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-5090": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "RTX 5090", "vcpu": 7, "ram_gib": 64, "disk_gib": 50, "hour": 0.8775, "month_cap": null } ] }, { "key": "gpu-rtx-4000-ada", "label": "RTX 4000 Ada live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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-4000-Ada": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension. Added 2026-10-14: offers API \"gpu\":\"RTX4000Ada\",\"hourly_micros\":1066500, node 8000 millicores / 32768 MB, available_nodes 1; homepage \"RTX4000Ada — 8 32 GB 1 ... $1.07 /hr\".", "sizes": [ { "name": "RTX 4000 Ada", "vcpu": 8, "ram_gib": 32, "disk_gib": 50, "hour": 1.0665, "month_cap": null } ] }, { "key": "gpu-a-4000", "label": "A 4000 live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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-A4000": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "A 4000", "vcpu": 8, "ram_gib": 45, "disk_gib": 50, "hour": 1.08, "month_cap": null } ] }, { "key": "gpu-l-40", "label": "L 40 live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": { "L40": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "L 40", "vcpu": 26, "ram_gib": 192, "disk_gib": 50, "hour": 1.188, "month_cap": null } ] }, { "key": "gpu-l-40-s", "label": "L 40 S live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "L 40 S", "vcpu": 12, "ram_gib": 72, "disk_gib": 50, "hour": 1.188, "month_cap": null } ] }, { "key": "gpu-l-4", "label": "L 4 live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": { "L4": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "L 4", "vcpu": 8, "ram_gib": 48, "disk_gib": 50, "hour": 1.2825, "month_cap": null } ] }, { "key": "gpu-rtx-6000-ada", "label": "RTX 6000 Ada live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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-6000-Ada": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "RTX 6000 Ada", "vcpu": 12, "ram_gib": 80, "disk_gib": 50, "hour": 1.3095, "month_cap": null } ] }, { "key": "gpu-a-10", "label": "A 10 live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": { "A10": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "A 10", "vcpu": 30, "ram_gib": 200, "disk_gib": 50, "hour": 1.7415, "month_cap": null } ] }, { "key": "gpu-a-100-80-gb", "label": "A 100-80 GB live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension. 2026-10-14 verify: offers API \"gpu\":\"A100-80\",\"hourly_micros\":1863000, node 14000 millicores / 102400 MB; homepage \"A100 80GB 80 GB 14 100 GB 1 ... $1.86 /hr\" (was $1.8225 on 28 vCPU / 120 GiB).", "sizes": [ { "name": "A 100-80 GB", "vcpu": 14, "ram_gib": 100, "disk_gib": 50, "hour": 1.863, "month_cap": null } ] }, { "key": "gpu-a-5000", "label": "A 5000 live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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-A5000": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "A 5000", "vcpu": 8, "ram_gib": 45, "disk_gib": 50, "hour": 1.917, "month_cap": null } ] }, { "key": "gpu-a-40", "label": "A 40 live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": { "A40": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "A 40", "vcpu": 24, "ram_gib": 120, "disk_gib": 50, "hour": 2.511, "month_cap": null } ] }, { "key": "gpu-a-100-40-gb", "label": "A 100-40 GB live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": { "A100-40GB": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "A 100-40 GB", "vcpu": 30, "ram_gib": 200, "disk_gib": 50, "hour": 2.6865, "month_cap": null } ] }, { "key": "gpu-rtx-pro-6000", "label": "RTX-PRO-6000 live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "RTX-PRO-6000", "vcpu": 16, "ram_gib": 192, "disk_gib": 50, "hour": 2.9565, "month_cap": null } ] }, { "key": "gpu-v-100-16-gb", "label": "V 100-16 GB live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": { "V100": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "V 100-16 GB", "vcpu": 8, "ram_gib": 30, "disk_gib": 50, "hour": 3.159, "month_cap": null } ] }, { "key": "gpu-v-100-32-gb", "label": "V 100-32 GB live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": { "V100-32GB": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension.", "sizes": [ { "name": "V 100-32 GB", "vcpu": 8, "ram_gib": 30, "disk_gib": 50, "hour": 3.159, "month_cap": null } ] }, { "key": "gpu-h-100", "label": "H 100 live bundle", "flags": [], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": 0 }, "note": "GPU included in bundle price; minimum 1 physical cards. Whole-node price must not be multiplied again; card count/resource enforcement needs an engine extension. 2026-10-14 verify: offers API \"gpu\":\"H100\",\"hourly_micros\":8086500, node 16000 millicores / 274432 MB; homepage rate card \"H100 80 GB 16 268 GB 1 ... $8.09 /hr\" (was $3.6855 on 20 vCPU / 128 GiB).", "sizes": [ { "name": "H 100", "vcpu": 16, "ram_gib": 268, "disk_gib": 50, "hour": 8.0865, "month_cap": null } ] }, { "key": "gpu-h-200", "label": "H 200 live bundle", "flags": [ "sales", "stock" ], "os": [ "linux" ], "regions": null, "gpu_only": true, "features": { "isolation": "vm" }, "pricing": "sizes", "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": 5.4 }, "note": "Sold only as an 8-card box: $43.20/h per box (208 vCPU / 1904 GiB) = $5.40 per card-hour; encoded as gpu H200 $5.40 x gpu_min_count 8 with a $0 base, so 1-GPU requests are priced as the whole 8-card box and shown as a compromise. Not in api.tama.computer/v1/offers on 2026-09-28 (stock-limited). llms.txt: --gpu-count >1 relies on multi-card boxes where min gpus says 8.", "sizes": [ { "name": "H 200", "vcpu": 208, "ram_gib": 1904, "disk_gib": 50, "hour": 0, "month_cap": null } ], "gpu_min_count": 8, "stock_note": "H200 absent from the live offers API on 2026-09-28" } ], "storage": { "included_disk_gib": 50, "disk_gib_month": null, "disk_billed_when_stopped": false, "snapshot_gib_month": 0, "paused_ram_billed": false }, "network": { "egress_free_gib": null, "egress_gib": null, "ipv4_month": null, "ipv4_available": null }, "plans": [ { "name": "Pay per running second", "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": "No published subscription fee or concurrency guarantee; current offers vary." } ], "free": { "monthly_credit": null, "one_time_credit": null }, "features": { "linux": true, "windows": false, "macos": false, "arm64": false, "gpu": true, "snapshot": "mem", "pause_resume": true, "fork": true, "persistent_disk": true, "volumes": null, "auto_stop_idle": true, "max_session_h": null, "custom_image": true, "docker_inside": true, "nested_virt": null, "root": null, "systemd": null, "browser": false, "desktop": true, "computer_use": false, "code_interpreter": null, "agent_harnesses": true, "ssh": true, "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": false, "byoc": false, "open_source": false, "soc2": false, "hipaa": false, "sso": false, "gpu_types": [ "RTX4090", "A6000", "RTX5090", "RTX4000Ada", "A4000", "L40", "L40S", "L4", "RTX6000Ada", "A10", "A100-80GB", "A5000", "A40", "A100-40GB", "RTX-PRO-6000", "V100-16GB", "V100-32GB", "H100", "H200" ], "regions": null, "lang_python": true, "lang_node": true, "lang_go": null, "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": true, "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": false, "isolation": "vm", "image_oci": true, "snapshot_auto": true, "snapshot_on_demand": true, "harness_api": true, "ingress_rules": true, "secret_proxy": false, "anti_bot": false, "captcha_solving": false, "volume_attach": false, "volume_shared": false, "own_agent_api": false, "guest_firewall": true, "max_session_unlimited": true, "egress_ip_type": "datacenter", "usage_api": false, "persist_scope": "folder", "persist_path": "/workspace" }, "perf": { "cpu_runs_s": null, "cold_start_ms": null, "burst_tti_ms": null, "dax_s": null }, "caveats": [ "Storage, snapshot GiB-month and public IPv 4 prices are not published in the reviewed sources; null is not free.", "GPU prices are live offers and drift between checks: on 2026-10-14 the offers API and homepage rate card agree on A6000 $0.77 (6 vCPU / 48 GB), A100 80GB $1.86 (14 / 100 GB), A100 40 GB $2.69 (30 / 200 GB) and H100 $8.09 (16 / 268 GB).", "Whole-node rates include multiple cards; do not multiplyH 200$43.20 by 8 again. Docs initially say checkpointing only one card, while offers list multi-card minima:capability/launch conflict requires confirmation.", "Default idle timeout is disabled;--ttl minimum 60 seconds is opt-in. Files only persist under /workspace. Warm resumes retain RAM; fork descriptions guarantee baseline disk, not all process memory.", "Homepage calculator says $20 ofcredit buys 220 hours; that is not proof of a free $20 signup grant. Network pricing, concurrency andsnapshot retention unverified.", "Desktop: llms.txt documents an optional virtual desktop (`tama new --desktop`, \"these machines resume cold\"; SDK tama.desktop() returns a browser desktop URL) and a browser terminal. No managed browser/CDP endpoint or computer-use (screenshot/click) API is documented, so browser and computer_use are false.", "Egress and extra-disk prices: still not published (llms.txt / docs / offers API re-checked 2026-09-28; /workspace is a fixed durable 50 GiB disk, no resize price). Left null.", "Live offers drift: on 2026-09-28 the offers API briefly showed H100 $3.375/h and L40S $2.295/h; on 2026-09-29 it is back to H100 $3.6855 and L40S $1.188 (= card). H200/B200/Gaudi 2 are still absent from the live API and live rate card (only in the site's static fallback catalogue). V100/V100-32 ($3.159/h) are live offers, not discontinued. On 2026-10-14 A16 is gone from both the offers API and the homepage rate card (mode removed), RTX4000Ada ($1.0665/h) appeared, and H100 jumped to $8.0865/h (hourly_micros 8086500).", "GPU machines are one card each: '--gpu-count' >1 fails at create because checkpointing more than one card is unsupported. Multi-card nodes are billed whole (homepage: 'where that node carries more than one card, that is the fewest you rent'); no multi-card-minimum row is live on 2026-10-14. Whole-node rentals are 'billed whole for as long as your machine holds it' and may need 'minutes' to provision; serverless-pool cards are per card-hour on shared capacity. --capacity auto|serverless|on-demand exists but the offers API returns one price per card.", "Offers API 2026-09-29: RTX4090 and RTX5090 rows have available_nodes 0 and node size 0 (probably serverless-pool cards with no warm workers, not sold out); an A10G row exists with hourly_micros 0 and cards_on_node 0 (placeholder, not orderable). llms.txt also names B200, H200, T4, GH200, B300 as cards, but 'tama offers is the list that is true today' and none of these are in it.", "StarSling's 2026-09-29 benchmark run detected an OCI container (not a VM) inside this provider's 4 vCPU / 8 GB sandbox; the card keeps the provider's documented isolation (vm) until confirmed.", "Offers API 2026-10-14 names the A100 40 GB row plain \"A100\" (hourly_micros 2686500, same as the card) and RTX4090 still shows available_nodes 0." ], "sources": [ "https://tama.computer/", "https://tama.computer/llms.txt", "https://tama.computer/docs", "https://api.tama.computer/v1/offers" ], "audit_2026_09_29": { "applied": 27, "source": "checked against the provider's live pricing and docs pages", "reviewed": 2 }, "feature_evidence": { "snapshot_auto": { "value": true, "source": "https://tama.computer/llms.txt", "quote": "a stopped machine is billed for\nnothing, and stopping snapshots it on the way down.", "note": "Triggers: every stop (API/CLI/dashboard) unless --no-snapshot, and the opt-in idle policy (--ttl / auto_stop_seconds, min 60 s, disabled by default: 'Tama snapshots and uses the ordinary stop path when the deadline expires'). Memory included ('warm') for normal machines: 'A warm restore comes back in about a second with the processes that were running still running, GPU memory and CUDA context included.' Machines created with --desktop capture disk only (cold). Also automatic crash-recovery snapshots ('snapshot list --all: Include the ones taken automatically to recover from a crash'). Start restores automatically.", "checked": "2026-09-29" }, "snapshot_on_demand": { "value": true, "source": "https://tama.computer/docs", "quote": "Capture a snapshot without stopping", "note": "tama snapshot create / tama.snapshots.create(machine, label) on a running machine; snapshot list shows RESTORES 'memory+disk' or 'disk'. New machines boot from it: tama fork and templates (private or public). But 'Memory checkpoints belong to stop and start on one machine; a fork or template launch always boots the exact filesystem cold.' So memory is kept only for restoring the same machine; forks/templates are disk-only (template has_memory flag exists for same-workspace launches per llms.txt).", "checked": "2026-09-29" }, "harness_api": { "value": true, "source": "https://tama.computer/docs", "quote": "A thin launcher starts the existing Claude or Codex CLI inside the machine; tama does not replace either provider's harness, tools, instructions or subagents. The server owns a machine-scoped AgentSession and copies stdout and stderr into its durable transcript.", "note": "Harnesses: Claude Code and Codex (both in base image). CLI tama prompt / SDK prompt(name, prompt, agent='claude'|'codex', detach=True) returns a session id; tama logs replays/follows the durable transcript; tama.sessions.list(machine) lists sessions; dashboard Agents tab. No documented follow-up/resume-conversation call (each prompt is a new headless run); stream is raw stdout/stderr, not structured events.", "checked": "2026-09-29" }, "ingress_rules": { "value": true, "source": "https://tama.computer/docs", "quote": "The URL is the credential. Its hostname starts with a forty-character random machine token and ends with the port, so it will not be guessed and stays unique across workspaces. Anyone you send it to can reach the port. Treat it as a shared secret rather than as an address, and use tama port-forward when it should not leave your machine.", "note": "Only ports you publish (tama expose/unexpose, --port) get a public unguessable HTTPS URL; no auth/IP allowlist on published URLs. Private alternative: tama port-forward over the authenticated tunnel. 'There is no SSH port open to the internet' (SSH via ProxyCommand tama tunnel). Ports 2222/7681/8080 reserved. Weak control (bearer-URL), but per-port choice + private forwarding.", "checked": "2026-09-29" }, "secret_proxy": { "value": false, "source": "https://tama.computer/docs", "quote": "Inject a workspace secret as that environment variable", "note": "Workspace secrets are write-only in the API but are injected as env vars into the machine / agent process ('the encrypted runtime store, which injects it only into that agent process'); they are not in snapshots/forks/templates, but the sandboxed process sees the real value. No credential-injecting proxy.", "checked": "2026-09-29" }, "anti_bot": { "value": false, "source": "https://tama.computer/", "quote": "Give your agent a machine to hack on and pay by the second. Use it for coding agents, kernel development, or inference optimization.", "note": "not offered: outside the product's scope (a general-purpose Linux VM/GPU compute product, not a browser-automation/scraping product; no stealth or fingerprinting feature is documented)", "checked": "2026-09-30" }, "captcha_solving": { "value": false, "source": "https://tama.computer/", "quote": "Give your agent a machine to hack on and pay by the second. Use it for coding agents, kernel development, or inference optimization.", "note": "not offered: outside the product's scope (compute/GPU VM product with no browser/CAPTCHA feature documented)", "checked": "2026-09-30" }, "wake_on_request": { "value": false, "source": "https://tama.computer/docs", "quote": "Auto-stop is disabled by default. Set auto_stop_seconds=900 when creating a machine to snapshot and stop it after 15 minutes without Tama-visible activity.", "note": "Stopped machines are restarted only via explicit tama start/machine.start() calls; no documented behavior where an inbound HTTP/SSH request to a stopped machine automatically wakes it.", "checked": "2026-09-30" }, "volume_attach": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "Work under `/workspace`. It is the durable 50 GiB disk, every shell starts there, and it is the only place edits survive a stop.", "note": "Only one fixed durable disk (/workspace) is documented; no CLI/SDK command creates, attaches, or detaches additional volumes.", "checked": "2026-09-30" }, "volume_shared": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "Work under `/workspace`. It is the durable 50 GiB disk, every shell starts there, and it is the only place edits survive a stop.", "note": "Each machine has its own private disk; forks copy the disk rather than sharing it ('A fork boots cold from a copy of the snapshot's disk ... no attempt inherits anything from another'). No shared-volume feature documented.", "checked": "2026-09-30" }, "live_resize": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "--cpu fractional allowed; 1 is the floor\n--memory MiB, not GiB; 1024 is the floor", "note": "CPU/RAM are set only at machine creation (--cpu/--memory flags); no documented API/CLI call to change resources of a running machine without recreating it.", "checked": "2026-09-30" }, "own_agent_api": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "A thin launcher starts the existing Claude or Codex CLI inside the machine; tama does not replace either provider's harness, tools, instructions or subagents.", "note": "not offered: outside the product's scope. tama only launches third-party coding-agent CLIs (Claude Code, Codex) inside the VM (harness_api); it has no first-party browser/computer-use/research agent of its own.", "checked": "2026-09-30" }, "mcp_server": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "Every flag, both SDK quickstarts, and the lines each command prints are at https://tama.computer/docs", "note": "No MCP server is mentioned anywhere in the CLI docs, llms.txt, or Python/TypeScript SDK guides; control surfaces are the CLI and the two SDKs (gRPC/Connect) only.", "checked": "2026-09-30" }, "lang_python": { "value": true, "source": "https://tama.computer/llms.txt", "quote": "tama exec box python train.py # streams output, exits with its code", "note": "Base image ships Python (used directly in worked tasks: pip install, python train.py, pytest -q) as well as Node/npm.", "checked": "2026-09-30" }, "public_ipv4": { "value": false, "source": "https://tama.computer/docs", "quote": "There is no SSH port open to the internet. The generated config sets ProxyCommand to tama tunnel, so the connection is carried over the same authenticated channel as everything else and port 2222 stays inside the machine.", "note": "Inbound reachability is only via provider-generated HTTPS URLs for explicitly exposed ports (tama expose) or the authenticated tunnel/port-forward; no raw public IPv4 address is assigned to or exposed for a machine.", "checked": "2026-09-30" }, "egress_allowlist": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "Every flag, both SDK quickstarts, and the lines each command prints are at https://tama.computer/docs", "note": "not documented: no outbound allow/deny-list, no-internet mode, or security-group egress control is described anywhere in the CLI/SDK docs.", "checked": "2026-09-30" }, "egress_http_rules": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "Every flag, both SDK quickstarts, and the lines each command prints are at https://tama.computer/docs", "note": "not documented: no method/path-level egress control is described anywhere in the CLI/SDK docs.", "checked": "2026-09-30" }, "guest_firewall": { "value": true, "source": "https://tama.computer/llms.txt", "quote": "A machine is a Linux VM you can SSH into, run an agent inside, snapshot mid-task, fork and stop.", "note": "derived: full Linux VM with Docker support inside (docker_inside=true) implies guest-side nftables/iptables usage is possible; not separately documented.", "checked": "2026-09-30" }, "windows": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "A machine is a Linux VM you can SSH into, run an agent inside, snapshot mid-task, fork and stop.", "note": "not offered: outside the product's scope (only Linux/Ubuntu machines are described; no Windows image or flag exists)", "checked": "2026-09-30" }, "macos": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "A machine is a Linux VM you can SSH into, run an agent inside, snapshot mid-task, fork and stop.", "note": "not offered: outside the product's scope (only Linux/Ubuntu machines are described; no macOS/Apple-silicon VM offering exists)", "checked": "2026-09-30" }, "arm64": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "Cards include B200, H200, H100, A100-80, RTXPro6000, L40S, RTX5090, A6000, RTX4090, and others such as T4, L4, A10G, V100, GH200 and B300.", "note": "not offered: outside the product's scope. All documented machine/card options are x86 (Ubuntu base image, NVIDIA GPU nodes); no ARM/aarch64 CPU option is mentioned.", "checked": "2026-09-30" }, "self_host": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "`--api` and `--console` point the CLI somewhere other than production, and fall back to `TAMA_API` and `TAMA_CONSOLE`.", "note": "tama is a hosted SaaS control plane/gateway (production API/console); no self-hosting instructions or on-prem deployment option are documented.", "checked": "2026-09-30" }, "byoc": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "`--api` and `--console` point the CLI somewhere other than production, and fall back to `TAMA_API` and `TAMA_CONSOLE`.", "note": "No bring-your-own-cloud/deploy-into-your-account option is documented; the service runs machines on tama's own infrastructure via its production gateway.", "checked": "2026-09-30" }, "open_source": { "value": false, "source": "https://github.com/tama-computer", "quote": "Not Found", "note": "No public GitHub org/source repository for the core tama service was found; the Python/TypeScript packages are client SDKs, not the core sandbox product, and no OSI license is claimed for the platform itself.", "checked": "2026-09-30" }, "soc2": { "value": false, "source": "https://tama.computer/security", "quote": "URL: https://tama.computer/security STATUS: 200 (redirects to the homepage; no dedicated security/trust content page exists)", "note": "not offered: no security/compliance/trust page or SOC 2 mention exists on the site or in docs.", "checked": "2026-09-30" }, "hipaa": { "value": false, "source": "https://tama.computer/trust", "quote": "URL: https://tama.computer/trust STATUS: 200 (redirects to the homepage; no dedicated trust/compliance content page exists)", "note": "not offered: no HIPAA/BAA mention exists anywhere on the site or in docs.", "checked": "2026-09-30" }, "sso": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "`tama login` opens a browser, writes the credential to `~/.tama/config.yaml`, and generates an SSH key and config of its own.", "note": "Only a single browser-based login / API token flow is documented; no SAML/OIDC SSO feature is mentioned.", "checked": "2026-09-30" }, "eu_data_residency": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "`--api` and `--console` point the CLI somewhere other than production, and fall back to `TAMA_API` and `TAMA_CONSOLE`.", "note": "No region/location selection flag or EU-residency option is documented anywhere (card's own `regions` field is also null); machine placement/region is not exposed to users.", "checked": "2026-09-30" }, "max_session_h": { "value": "none", "source": "https://tama.computer/llms.txt", "quote": "You are billed for the seconds a machine is running; a stopped machine is billed for nothing, and stopping snapshots it on the way down.", "note": "A machine (VM) runs until explicitly stopped/rm'd or until an opt-in idle auto-stop (--ttl, disabled by default) fires; no maximum wall-clock session length is imposed.", "checked": "2026-09-30" }, "memory_snapshot_fork": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "A fork boots cold from a copy of the snapshot's disk, which is what makes attempts comparable: no attempt inherits anything from another, and neither can write over the other.", "note": "Docs explicitly say fork is disk-only/cold; memory ('warm', with running processes/GPU context) is only restored via stop+start on the SAME machine, not carried into a fork. Confirms false.", "checked": "2026-09-30" }, "browser": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "tama.desktop(name_or_id) # browser desktop URL\n tama.terminal(name_or_id) # browser terminal URL", "note": "The only browser-related surfaces are a browser-viewable virtual desktop (`--desktop`, VNC-style stream) and a browser-viewable SSH terminal — both are the human viewing a remote UI over HTTP, not a managed/automatable browser instance. No headless browser, CDP endpoint, or browser-as-a-service is documented anywhere in llms.txt or /docs. Confirms false.", "checked": "2026-09-30" }, "computer_use": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "`--desktop` include the virtual desktop; these machines resume cold", "note": "The `--desktop` flag adds a VNC-like virtual desktop reachable via a browser URL (tama.desktop()), but there is no documented API to drive it programmatically by screenshots/clicks (no screenshot/click/type endpoints in the CLI, Python SDK, TypeScript SDK, or docs). desktop=true is correct (graphical desktop exists) but computer_use (screenshot+click driving API) is not offered. Confirms false.", "checked": "2026-09-30" }, "egress_ip_type": { "value": "datacenter", "source": "https://tama.computer/", "quote": "Give your agent a machine to hack on and pay by the second. Use it for coding agents, kernel development, or inference optimization.", "note": "derived: tama is a general-purpose Linux VM/GPU compute provider (not a proxy/scraping/residential-IP product); no residential, mobile, or ISP-IP feature is documented anywhere in llms.txt, /docs, or the offers API, so outbound traffic is treated as ordinary datacenter IP space typical of this class of provider.", "checked": "2026-09-30" }, "static_egress_ip": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "There is no SSH port open to the internet. The generated config sets ProxyCommand to tama tunnel, so the connection is carried over the same authenticated channel as everything else and port 2222 stays inside the machine.", "note": "No IP-address feature of any kind (public IPv4, fixed/static outbound IP, purchasable IPv4) is mentioned anywhere in llms.txt, /docs, or the pricing page; the network section of the card (ipv4_available, ipv4_month) is also null/unpublished. No fixed outbound IP option is documented.", "checked": "2026-09-30" }, "usage_api": { "value": false, "source": "https://tama.computer/llms.txt", "quote": "Complete Tama surface: tama.usage(days=30) # balance and metered usage", "note": "granularity: account-only; cost: yes; The tama.usage(days=30) method in Python SDK and await tama.usage(30) in TypeScript SDK returns balance and metered usage, but only at the account level with a days parameter. There is no documented way to query usage for a specific machine/sandbox by ID, name, tag, label, or project. The method signature does not accept a machine identifier, making it account-only usage reporting rather than per-sandbox or per-group usage.", "confidence": "medium", "checked": "2026-10-01" }, "persist_scope": { "value": "folder", "source": "https://tama.computer/llms.txt", "quote": "a stopped machine is billed for\nnothing, and stopping snapshots it on the way down.", "note": "Only /workspace survives a stop; the rest of the filesystem (packages installed elsewhere) resets.", "checked": "2026-10-01" } }, "locations": [], "locations_source": "https://tama.computer/docs", "locations_quote": "No region/location selection flag or EU-residency option is documented; machine placement/region is not exposed to users.", "locations_checked": "2026-10-01" }