What is Claude Managed Agents?
Hosted agent sessions with memory and orchestration on Anthropic's platform.
Who is Claude Managed Agents best for?
Claude Managed Agents is best for developers who want Anthropic to host the loop and sandbox for long agent jobs.
Long-running agent jobs
Sessions can run for minutes or hours with many tool calls, and runtime is billed only while the agent is actually working.
Teams without agent infrastructure
Anthropic hosts the agent loop, sandbox and tool execution, so you skip building and running your own.
Jobs split across several agents
A lead agent can hand parts of a task to specialist agents, each with its own model, prompt and tools, working on a shared filesystem.
Work that must meet a set standard
With outcomes you write a rubric, and a separate grader checks the result and sends the agent back for another pass.
How much does Claude Managed Agents cost?
Claude Managed Agents charges $0.08 per session-hour of active runtime plus standard Claude API token rates, which start at $1 per million input tokens and $5 per million output tokens on Haiku 4.5.
- Billed only while a session is running, not while idle
- Measured to the millisecond
- Replaces container-hour billing for code execution
- Cache reads $0.10, cache writes $1.25
- Anthropic's fastest, lowest-cost model
- Cache reads $0.20, cache writes $2.50
- Cache reads $0.20, cache writes $5
- Fast mode at 2x the standard price
Fable 5.1 tokens cost $10 / $50 per million input / output tokens. Web searches inside a session cost $10 per 1,000. Pinning an agent to US-only inference bills its tokens at 1.1x, and the Batch API's 50% discount does not apply because sessions are interactive. Anthropic's own example: a one-hour session on Opus 5 using 50,000 input and 15,000 output tokens costs about $0.71. Cache write prices shown are for the 5-minute cache. Full pricing on claude.com. We checked these prices on .
How do readers rate Claude Managed Agents?
No one has rated Claude Managed Agents yet. Be the first.
Sign in to rate. One rating per account.
What are the best Claude Managed Agents alternatives?
Three tools on Recaply do Claude Managed Agents's job. All three are open source, so you can run them on your own server.
AgnoOpen SourceChoose it when you want agents served as APIs and MCP servers from your own cloud, with any model provider rather than one
eveOpen SourceChoose it when your TypeScript team wants to build durable backend agents with Vercel's Apache-licensed framework and run them on your own infrastructure.
NemoClawOpen SourceChoose it when you want a free sandbox that controls the credentials and network access of agents you run yourself.
More open source alternatives
See all
Questions about Claude Managed Agents
What is Claude Managed Agents?
Hosted agent sessions with memory and orchestration on Anthropic's platform.
Who is Claude Managed Agents for?
Claude Managed Agents is best for developers who want Anthropic to host the loop and sandbox for long agent jobs.
Is Claude Managed Agents free?
No. Claude Managed Agents is a paid tool, from $0.08 per session-hour.
How much does Claude Managed Agents cost?
Claude Managed Agents's paid plans are Session runtime at $0.08 per session-hour, Haiku 4.5 tokens at $1 / $5 per million input / output tokens, Sonnet 5 tokens at $2 / $10 per million input / output tokens and Opus 5.5 tokens at $4 / $20 per million input / output tokens.
Is there an open source alternative to Claude Managed Agents?
Yes. Agno, eve, NemoClaw and Omnara are open source tools that do the same job, so you can run your own.
What are the best Claude Managed Agents alternatives?
The closest alternatives on Recaply are Agno, eve and NemoClaw.
Where can I try Claude Managed Agents?
On its official website, platform.claude.com.
Does Claude Managed Agents charge for idle sessions?
No. The $0.08 per session-hour runtime fee counts only while a session is running, measured to the millisecond. Time spent idle, such as waiting for your next message or a tool confirmation, is not billed, though tokens are charged at normal model rates whenever the agent works.
Does the Batch API discount apply to Claude Managed Agents?
No. Managed Agents sessions are stateful and interactive, so there is no batch mode and no 50% batch discount. Prompt caching discounts still apply, and pinning an agent to US-only inference adds 10% to its token prices.
Claude Platform
Open Memory Protocol
DigitalOcean Managed Agents
Claude Code Agent View
Xirp
Enterprise-Managed Authorization