OpenClaw
Hermes
Claude Code
CodexT1
In-runtime repair
Restart with bounded backoff, then run the runtime doctor when a configuration can be repaired in place.
Always-on AI agent hosting Run your agent 24/7.
Your agent keeps working day and night, even with your laptop closed.
From$2.88a month per agent
Billed to the dayNo network fees
Early access: book a short call and we open your account.
- Any harnessOpenClaw, Hermes, Claude Code, Codex or your own
- Repairs itselfIf it breaks, Molted fixes it at no extra cost
- 1,500+ toolsGmail, Slack, Notion, HubSpot and more
- Your own modelYour Claude Code or Codex subscription, or any API key
One agent, always onIllustration
1.1s
OpenClaw cold start
<16s
to provision
2×
faster per core than a cloud VM
2 regions
EU · North America
01Performance benchmark
Cold start
in 1.1 seconds.
We benchmark the runtime itself and keep the setup that gets an agent ready fastest.
Runtime performance lab
Cold-start trial
Mac miniHardware price · $799
5.1sDesktop baseline
MacBook Pro M3 ProHardware price · $3,400
4.0sHigh-end Mac baseline
Molted premiumBenchmarked for agents
1.1sBest measured result
faster than the high-end Mac benchmark.Median of repeated internal trials · OpenClaw cold start · representative hardware
01.1sAfter the benchmark, the operating layer keeps the advantage useful.
SelectionRuntime-tested setup
RegionsEU and North America
Instance isolationPer-instance resource limits
Why not a bigger VPS?
Be aware of regular hosts selling agent hosting.
Their offer is often a standard VPS with more RAM for the price. An agent works one step at a time and each step runs on one core. More RAM lets it run more tasks in parallel but does not make a step faster. A well-built runtime runs well in 3 to 6 GB.
2×faster per core than a standard cloud VM, on the same tasksInternal measurement, median of three runs: 1.7× to 2.1× depending on the task.
Agent steps done in the same minute
Bigger VPSMore RAM, slower cores
5 stepsMoltedLess RAM, faster cores
10 steps2× more- 01
Server processors are slower per core
Budget VPS plans run on server processors built for a very large number of cores. Each core is slower. An agent mostly moves forward on one core at a time.
- 02
Memory scans take processor time
To fit more customers on each machine, budget hosts often scan and merge duplicate memory around the clock. That work uses the same processors as your agent and adds latency.
- 03
We choose processors for speed per core
We test processors on real agent work and keep the fastest per core.
02Built to recover
When an agent fails, Molted repairs it on its own.
Each instance is monitored from inside its runtime and from the outside. When something fails, Molted captures the evidence, escalates through recovery tiers and restores service automatically.
00:00Process stops responding
00:05Heartbeat missed
00:12Post-mortem captured
00:47Instance healthy
Illustrative recovery trace
47sec
2
monitoring layers
4
recovery tiers
Auto
post-mortem capture
24/7
fleet monitoring
Every failed recovery produces a post-mortem and escalates with its diagnostic context intact.
State you can inspect and restore
Every file change leaves a trace.
Inspect what changed. Compare revisions. Restore a known-good workspace, even after a delete. Plain, portable files keep the agent's state independent of its compute.
AutomaticEvery workspace revision is captured
InspectableDiff files between any two points
PortablePlain files, no proprietary memory format
workspace / atlas-prod
savedFiles
▾ workspace
▾ memory/
findings.md
▾ skills/
customer-research.md
SOUL.md
rev 184 · 12s ago
customer-research.mdv183 → v184
# Customer research skill
- Summarize the first result
+ Cross-check primary sources
+ Flag conflicting evidence
Return findings with citations
Illustrative workspace diff · plain files, versioned
03Pricing
Pick a popular size or build your own.
Large and XLarge are our most popular sizes. For anything else, set your own from 1.5 to 12 GB of RAM at the same unit prices. Billing is prorated and your main model stays on your own key or subscription.
Large
Popular$5.25per month
$4.38 a month billed yearly, two months free
- Up to 3 vCPU
- Up to 3 GB of RAM
- 10 GB of NVMe storage
- Automatic repair
- Versioned workspace and file explorer
One agent that mostly calls APIs: Hermes, Claude Code, Codex or a light harness of your own.
XLarge
Popular$7.90per month
$6.58 a month billed yearly, two months free
- Up to 3 vCPU
- Up to 6 GB of RAM
- 20 GB of NVMe storage
- Automatic repair
- Versioned workspace and file explorer
OpenClaw, a browser that stays open, builds or several sub-agents at once.
Build your own
Move the sliders to price any size.
Hot resizing: change the size of a running agent, no restart.$5.25 per month
per month, always on
About $0.18 a day, prorated to the time your agent exists.
| vCPU | 3 × $0.70 | $2.10 |
|---|---|---|
| RAM | 3 × $0.65 | $1.95 |
| NVMe storage | 10 × $0.07 | $0.70 |
| Per agent | flat | $0.50 |
- $0.53for 3 days of Large, because billing is prorated
- 10×cheaper while the agent is stopped, workspace kept
- $0once the agent is deleted
- $0network fees, in or out
Included in every size
- Always onYour runtime installed, kept up to date and never put to sleep.
- Automatic repairMonitoring and four recovery tiers bring your agent back with its state when something breaks.
- Hot resizingChange the size of a running agent without a restart.
- Versioned workspaceEvery file and every config change keeps its history. Compare two versions side by side and restore the one you want.
- File explorerBrowse, search, edit and download your agent's files from the dashboard.
- Console, SSH, API and URLA terminal in your browser, SSH access, the API and a public HTTPS address for every agent.
- Logs and metricsLive logs, resource usage and a timeline of every event on your instance.
- NVMe storageEvery disk is NVMe SSD, never a spinning hard drive.
On demand, billed on usage
Ready from day one, with no extra API key.
Your main model stays on your own key or subscription. Everything below works without one: you pay only for what your agent uses.- Stealth browserA cloud browser that solves captchas on its own and can browse from residential IPs.
- 1,500+ toolsGmail, Slack, Notion, HubSpot and more than 1,500 other apps your agent connects to by itself.See all integrations
- Long-term memoryMemory search with reranking, so your agent pulls the right context out of everything it has stored.
- VoiceSpeech to text and text to speech, so your agent can listen and answer out loud.
- Decision modelsSmall, fast models for routing and structured choices, so your main model is kept for the real work.
04Compare
What the same agent costs elsewhere.
Public prices for one agent running all month, at the same size. Sort them by price, per vCPU or per GB of RAM.
Offer
Size compared
vs Molted
Billing
MoltedLarge / XLarge
Size compared3 vCPU · 3 GB · 10 GBResources at unit price, plus $0.50 per agent
Per month, always on$5.25Lowest
vs Molted1×
Per vCPU$0.70
Per GB RAM$0.65
BillingProrated, no network feesSleeps: never
Agent37Cloud sandbox
Size compared3 vCPU · 3 GB · 10 GB
Per month, always on$5.40
vs Molted1.03×
Per vCPU$0.80
Per GB RAM$0.70
BillingMetered, prepaidSleeps: optional
Agent37Cloud VM
Size compared3 vCPU · 3 GB · 10 GB
Per month, always on$12.15
vs Molted2.3×
Per vCPU$2.00
Per GB RAM$1.75
BillingMetered, prepaidSleeps: optional
InstaboxPro
Size compared2 vCPU · 4 GiB, closest size$20 plan minimum, covers about $14 of usage
Per month, always on$20.00
vs Molted3.8×
Per vCPU$2.92
Per GB RAM$1.64
BillingPlan + usageSleeps: optional
DigitalOceanBasic droplet
Size compared2 vCPU · 4 GiB · 80 GiB, closest sizeUpdates, restarts and monitoring are yours
Per month, always on$24.00
vs Molted4.6×
Per vCPUBundle
Per GB RAMBundle
BillingFlat monthlySleeps: never
OrgoStarter
Size compared1 computer · 8 GB · 40 GB$24.17 billed yearly
Per month, always on$29.00
vs Molted5.5×
Per vCPU$5.00
Per GB RAM$2.50
BillingFlat monthlySleeps: never
machine0large / xl
Size compared2 vCPU · 4 GB · 80 GB, closest size
Per month, always on$37.96
vs Molted7.2×
Per vCPUBundle
Per GB RAMBundle
BillingPer minuteSleeps: optional
MaritimeStarter + always-on
Size compared1 vCPU · 3 GB · 10 GBThen $25 for each extra agent on the plan
Per month, always on$45.00
vs Molted8.6×
Per vCPUBundle
Per GB RAM$3.00
BillingPlan + add-onsSleeps: never, with the always-on add-on
Public list prices for one agent running 730 hours a month, in US dollars before taxes, checked on October 24, 2026. When an offer has no size equal to ours, we use its smallest size that covers the same memory. Model usage and promotional credits are excluded.
05Enterprise
Running a fleet or building a product on agents?
The same platform, operated for your customers or your teams, with the controls a larger deployment needs.
- Incident response by our engineers, 24/7
- Direct support on WhatsApp or Telegram
- An SLA on infrastructure availability
- On-premise, dedicated and sovereign deployments
01What does "up to 3 vCPU" mean?
Your agent can use all the vCPU and RAM of its size whenever it needs them. Agents spend most of their time waiting on models and tools, so they get the full ceiling when they work. For a dedicated deployment, see the Enterprise offer.
02What is included in the price?
Everything your agent needs to run all month: the always-on instance with your runtime installed and kept up to date, monitoring and automatic repair, a versioned workspace with a file explorer, a console in your browser, SSH access, the API, logs, metrics and its own HTTPS URL. Model usage is billed by your provider, with your own key or subscription.
03How is the price calculated?
Every size uses the same unit prices: $0.70 per vCPU, $0.65 per GB of RAM and $0.07 per GB of disk, plus a flat $0.50 per agent. Large: 3 vCPU, 3 GB of RAM and 10 GB of disk come to $4.75, so $5.25 with the agent. XLarge: 3 vCPU, 6 GB of RAM and 20 GB of disk come to $7.40, so $7.90 with the agent. Other sizes follow the same unit prices. Every price is prorated to the time the agent exists.
04Which size should I pick?
Large for one agent that mostly calls APIs. XLarge for OpenClaw, a browser that stays open, builds or several sub-agents. Those are our two most popular sizes. For a light script or a heavy workload, set your own size from 1.5 GB to 12 GB of RAM. You can change size at any time without restarting the agent. We also check real usage with you during the first week.
05What is billed on usage?
The stealth browser, more than 1,500 tools, long-term memory, voice and decision models. Your agent uses them without an API key of its own, you only pay for what it uses and you set a monthly spending cap.
06Do I pay for a full month if my agent runs a few days?
No. Billing is prorated: three days of Large cost about $0.53. A stopped agent costs 10× less, a deleted one costs nothing and network traffic is never billed.
07Which runtimes are supported?
OpenClaw and Hermes Agent, plus Claude Code or Codex through Mue, our runtime. If you run your own harness, we package it with you.
08Is there a yearly price?
Yes. Paying yearly gives you two months free: $4.38 a month for Large and $6.58 for XLarge.
09When should I use the Enterprise offer instead?
When customers or a whole company depend on your agents and you need commitments around them: incident response 24/7, direct support on WhatsApp or Telegram, an SLA with credits, on-premise or Swiss deployment. Enterprise runs the same platform and compute as self-serve, with that service on top. See the Enterprise page.
Self-serve · from $2.88 a month
Start with one agent.
Pick a size and bring your runtime. Change size whenever you need it, without a restart.