Launched on November 5, 2026, Kimi Hosted Agents is Moonshot AI’s managed 24/7 runtime for long running and multi agent work. The service is designed to spare teams from building and maintaining the agent runtime themselves, while offering isolated sandboxes and network controls.
Published byEdited with GPT-6 LunaImages generated with GPT Image 2
Research answer

Create a landscape editorial hero image for this Studio Global article: What is Moonshot AI’s Kimi Hosted Agents cloud service, reported on November 5, 2026, and how does its managed “Harness as a Service” approa. Article summary: Kimi Hosted Agents is Moonshot AI’s managed cloud service for the agent “harness”—the execution framework around a model—reported as launched on November 5, 2026. Its central proposition is a 24/7 hosted environment for . Topic tags: general, general web, documentation. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fak
Moonshot AI launched Kimi Hosted Agents on November 5, 2026, as a cloud service for the software layer that runs AI agents. Instead of building and maintaining that runtime themselves, teams can use a hosted environment intended for long-running tasks and collaboration between agents. The key pricing detail: sandbox time starts at ¥0.072 per hour, but model inference and plugin use are billed separately. 10
17
18
A model generates responses; an agent harness provides the surrounding execution environment—the session, tools and runtime needed to carry out a task. Kimi Hosted Agents packages that layer as a managed cloud offering, rather than requiring a team to operate it on its own infrastructure. Moonshot’s documentation describes a 24/7 environment for long-running tasks and multi-agent collaboration. 17
That distinction matters for teams building agents that need to do more than answer a single prompt. The offering is positioned as managed execution infrastructure around a model, not simply another way to call a model API. 17
Moonshot’s overview describes two broad use cases: internal agents for business monitoring, research and operational response, and agents embedded in customer-facing products—for example, investment-research assistants or in-game strategy advisers. 17
The documentation also describes isolated sandboxes and network-access controls, which let organizations set boundaries around agent activity. Launch coverage additionally reports that sessions can be stopped and audit records exported. 17
33
These are controls around execution, not evidence by themselves of a particular security certification or service-level guarantee. The available sources do not specify a runtime-upgrade schedule or uptime SLA. 17
33
The official pricing page breaks charges into three parts: sandbox runtime, model inference and plugin calls. Its formula is sandbox hours × ¥0.072 per hour, plus model inference fees and plugin charges. So ¥0.072 is the sandbox rate, not the total cost of running an agent. 18
Actual costs will depend on how long sessions run, how much model inference they use and how often they invoke chargeable plugins. Teams evaluating the service should account for all three components rather than comparing sandbox rates alone. 18
Launch reporting lists more than 140 skills and more than 30 plugins. These figures describe the reported catalog at launch; they do not, by themselves, establish that every integration or customization a particular team needs is available. 33
The clearest supported takeaway is that Moonshot is offering a managed agent runtime for teams that want persistent execution and multi-agent workflows without operating the harness themselves. The sources establish the core product proposition, examples of intended use, basic execution controls and the separate pricing components. They do not establish every deployment option or promise a specific maintenance cadence. 17
18
Studio Global AI
This page includes a source-backed answer you can continue inside Studio Global.
Launched on November 5, 2026, Kimi Hosted Agents is Moonshot AI’s managed 24/7 runtime for long running and multi agent work.
Launched on November 5, 2026, Kimi Hosted Agents is Moonshot AI’s managed 24/7 runtime for long running and multi agent work. The service is designed to spare teams from building and maintaining the agent runtime themselves, while offering isolated sandboxes and network controls.
Moonshot describes internal and customer facing uses, but the available sources do not specify an upgrade schedule or service level guarantee.
Launched on November 5, 2026, Kimi Hosted Agents is Moonshot AI’s managed 24/7 runtime for long running and multi agent work. The service is designed to spare teams from building and maintaining the agent runtime themselves, while offering isolated sandboxes and network controls.
Published byEdited with GPT-6 LunaImages generated with GPT Image 2
Research answer

Create a landscape editorial hero image for this Studio Global article: What is Moonshot AI’s Kimi Hosted Agents cloud service, reported on November 5, 2026, and how does its managed “Harness as a Service” approa. Article summary: Kimi Hosted Agents is Moonshot AI’s managed cloud service for the agent “harness”—the execution framework around a model—reported as launched on November 5, 2026. Its central proposition is a 24/7 hosted environment for . Topic tags: general, general web, documentation. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fak
Moonshot AI launched Kimi Hosted Agents on November 5, 2026, as a cloud service for the software layer that runs AI agents. Instead of building and maintaining that runtime themselves, teams can use a hosted environment intended for long-running tasks and collaboration between agents. The key pricing detail: sandbox time starts at ¥0.072 per hour, but model inference and plugin use are billed separately. 10
17
18
A model generates responses; an agent harness provides the surrounding execution environment—the session, tools and runtime needed to carry out a task. Kimi Hosted Agents packages that layer as a managed cloud offering, rather than requiring a team to operate it on its own infrastructure. Moonshot’s documentation describes a 24/7 environment for long-running tasks and multi-agent collaboration. 17
That distinction matters for teams building agents that need to do more than answer a single prompt. The offering is positioned as managed execution infrastructure around a model, not simply another way to call a model API. 17
Moonshot’s overview describes two broad use cases: internal agents for business monitoring, research and operational response, and agents embedded in customer-facing products—for example, investment-research assistants or in-game strategy advisers. 17
The documentation also describes isolated sandboxes and network-access controls, which let organizations set boundaries around agent activity. Launch coverage additionally reports that sessions can be stopped and audit records exported. 17
33
These are controls around execution, not evidence by themselves of a particular security certification or service-level guarantee. The available sources do not specify a runtime-upgrade schedule or uptime SLA. 17
33
The official pricing page breaks charges into three parts: sandbox runtime, model inference and plugin calls. Its formula is sandbox hours × ¥0.072 per hour, plus model inference fees and plugin charges. So ¥0.072 is the sandbox rate, not the total cost of running an agent. 18
Actual costs will depend on how long sessions run, how much model inference they use and how often they invoke chargeable plugins. Teams evaluating the service should account for all three components rather than comparing sandbox rates alone. 18
Launch reporting lists more than 140 skills and more than 30 plugins. These figures describe the reported catalog at launch; they do not, by themselves, establish that every integration or customization a particular team needs is available. 33
The clearest supported takeaway is that Moonshot is offering a managed agent runtime for teams that want persistent execution and multi-agent workflows without operating the harness themselves. The sources establish the core product proposition, examples of intended use, basic execution controls and the separate pricing components. They do not establish every deployment option or promise a specific maintenance cadence. 17
18
Studio Global AI
This page includes a source-backed answer you can continue inside Studio Global.
Launched on November 5, 2026, Kimi Hosted Agents is Moonshot AI’s managed 24/7 runtime for long running and multi agent work.
Launched on November 5, 2026, Kimi Hosted Agents is Moonshot AI’s managed 24/7 runtime for long running and multi agent work. The service is designed to spare teams from building and maintaining the agent runtime themselves, while offering isolated sandboxes and network controls.
Moonshot describes internal and customer facing uses, but the available sources do not specify an upgrade schedule or service level guarantee.