Compare · camelRun by camelAI

Bedrock AgentCore alternative

One meter for your AI agents

camelRun by camelAI runs your AI agents: the loop, their history and files, and calls to your tools.

camelRun by camelAI runs the agent loop on Bedrock, Anthropic, OpenAI, Google and other models, keeps each agent's history and files, and calls the tools in your own code. Agent time bills only while an agent is in a run.

Get an API key Read the quickstart

Moving off Amazon Bedrock Agents? Start here

agent.yaml

Define AI agents in YAML and deploy them with one command

Like an AgentCore harness, a camelRun agent starts as configuration: a model, instructions and tools in agent.yaml, in your repository. Run camelrun deploy and it's live. Deploy again with --apply, and live agents move to the new version between turns, keeping their history.

agent.yaml

1key: support                       # the same key is the same definition
2name: Support
3model: anthropic/claude-sonnet-5-5
4systemPromptFile: prompts/support.md
5builtins: [web_search, ask_user]
6mcpServers:
7  - name: app
8    url: https://app.example.com/mcp
9    auth: { type: bearer, token: "${APP_MCP_TOKEN}" }
10openApi:
11  - name: billing
12    specFile: openapi.yaml
13    auth: { type: runtime }
14agents:                            # keyed agents made from it on deploy
15  - key: support-main
16  - key: support-eu
17    systemPromptAppend: Answer customers in the EU region.

Terminal

$ camelrun deploy

support: created (def_4b1e09c7a2d65f38e910, revision 1)

tools from app: 4 tools

tools from billing: 6 tools

agent support-main: ready (client_9c2f7a1e04b83d6f5a0c1e9b72d48f3a6c05e1b7)

agent support-eu: ready (client_3e8a5d0c7f19b2e64a7d0c3f81b95e2a4d6c07f9)

$ camelrun run support-main "Which urgent tickets are about billing?"

One is about billing:

#5108: checkout keeps failing for a customer whose invoice is past due.

The CLI and agent.yaml

Pricing

Price a month of AI agents

Pick a workload. camelRun bills agent time while agents are in a run, and what they store, so idle and sleeping agents cost only storage.

AgentsA support agent for each customer

Minutes in a run, each day

History and files they keep

camelRun pricing Every AgentCore rate, with a worked example

camelRun · estimate30 days

Prepaid credit. No subscription, no seats and no per-agent fee.

camelRun pricing Every AgentCore rate, with a worked example

Amazon Bedrock models

Run your AI agents on Amazon Bedrock models

Add a Bedrock API key and its AWS region in the console, then point any agent at a Bedrock model such as Claude. Its history carries over.

Or over the API:

PUT /v1/providers/amazon-bedrock/keyModels and keys

Bedrock Agents (Classic)

A Bedrock Agents alternative: move your agents to camelRun

Amazon Bedrock Agents, now Bedrock Agents Classic, closed to new customers on July 30, 2026. Existing agents keep working, so you can move at your own pace, one agent at a time.

Bedrock Agents (Classic)agent.yaml and your code

Instructions4systemPromptFileThe same words, kept in your repository.
Foundation model3modelA Bedrock model, on your own Bedrock key.
Action group schema10openApiEvery operation becomes a tool, called at your API's URL.
Action group LambdacodeserveTools()Its functions, served from your own HTTPS endpoint, serverless included.
User confirmationyamlapprovalThe run waits for a yes, with nothing billed while it waits.
Multi-agent collaborationyamldelegateHand a task to another agent and get its answer back.

Tools from OpenAPI specs

Which to pick

AgentCore or camelRun: which fits your AI agents

AWS Bedrock AgentCore

Pick AgentCore when

Your team already runs on AWS, with IAM, VPCs and CloudTrail.
Your tools already run as Lambda functions, and you want AgentCore Gateway to serve them.
You wrote your own agent loop and want it hosted as it is.

camelRun by camelAI

Pick camelRun when

Each customer, ticket or thread gets its own agent, and thousands run at once.
You want tools in your own code, where the model never sees your credentials.
Your agents live in Slack, Discord, Telegram or GitHub, or wake on webhooks.
You want to self-host, or leave with all your data any time.

FAQ

Questions about AgentCore and camelRun

AWS Bedrock AgentCore

What is AWS Bedrock AgentCore?

AWS's platform for AI agents. Its Runtime hosts an agent loop you write, and its harness runs one for you from a model, a system prompt and tools, with Memory, Gateway and Observability around them, each billed on its own. camelRun by camelAI is closest to the harness.

Agent harness vs framework vs runtime

How is AgentCore priced?

By part: Runtime by vCPU-hour and GB-hour, Memory by GB and by record, Gateway by call and by tool indexed, Web Search by query and Observability at CloudWatch rates (AWS's pricing on October 22, 2026). The harness has no charge of its own, and Bedrock or your provider bills the tokens.

AWS pricing, checked October 22, 2026 Every AgentCore rate, with a worked example

What replaces Amazon Bedrock Agents for new accounts?

Bedrock Agents, now Bedrock Agents Classic, closed to new customers on July 30, 2026, and AWS points them to the AgentCore harness. On camelRun, the agent's instructions and Bedrock model go in agent.yaml, an action group's API becomes an OpenAPI tool source, and the functions its Lambda ran become tools served from your own code.

camelRun by camelAI

Can I use Amazon Bedrock models on camelRun?

Yes. Add a Bedrock API key and its AWS region on the console's Models & keys page, then name a model such as amazon-bedrock/global.anthropic.claude-sonnet-5-5. The usage bills to your AWS account.

Models and keys

Can I run AI agents outside AWS?

Yes. camelRun needs no AWS account: sign in at run.camelai.com and use models from Anthropic, OpenAI, Google and more, or run the open-source runtime yourself with one container and Postgres, on any cloud or your own servers.

Self-hosting camelRun

Does a camelRun agent remember past conversations?

Yes. Each agent has a key you choose, such as one per customer, and keeps its whole history and its files until you delete it.

Run AI agents on one meter

Sign in with GitHub or Google, create an API key, and deploy your first agent. Or paste one line into your coding agent, and it sets camelRun up for you.

Get an API key Read the quickstart

Set up with your coding agent

Read https://run.camelai.com/SKILL.md and set up camelRun in this project. Ask me what the agent should do.

Paste into Claude Code, Codex, Cursor or any coding agent. It asks what your agent should do, then builds it.