CrabTalk vs Hatchet.
Task orchestration + durable workflows
Hatchet routes tasks to workers and supplies orchestration, retries and durable workflow primitives. CrabTalk hosts bounded tool services with SQL state and per-call memory, CPU and row budgets.
Which fits your work?
CrabTalk
- Small shared operations with a SQL data boundary
- Calling or privately deploying existing service templates
- Hosted MCP access with per-tool resource limits
Hatchet
- Queued tasks and background work on your workers
- Durable tasks with sleeps and event waits
- Worker concurrency, routing and retry control
Use both.
A Hatchet task can call a CrabTalk HTTP tool to check a shared account rule before doing a larger import. Hatchet can own the import’s retries and concurrency without relocating existing application data.
Capabilities side by side.
| Capability | CrabTalk | Hatchet |
|---|---|---|
| Execution unit | A tool call in a service deployment | A task assigned to a registered worker |
| Code execution | Hosted Rust/Wasm instance per call | Worker processes in your chosen environment |
| Shared data | SQLite and key-value state per service | Task progress and your application’s chosen data store |
| Recovery | Transactions and scheduled wake retries | Retry policies and durable workflow primitives |
| Scheduling | Crons and one pending wake per service | Scheduled runs, crons, events and durable waits |
| Resource control | CPU, memory and row limits per tool | Worker slots, task slot costs and concurrency limits |
| Interfaces | HTTP and hosted MCP tools | SDK tasks and application triggers |
Check an import policy and process a large file.
Checking a small policy and parsing a large file need different resource budgets.
CrabTalk
- Reuse or deploy a policy service
- Expose the policy lookup with a small memory budget
- Call it before processing the file in another runtime
Hatchet
- Define policy and processing tasks
- Register them on workers sized for the file
- Configure concurrency, retries and run monitoring
CrabTalk vs Hatchet pricing.
CrabTalk
Callers pay a published tool’s price when it has one. Service owners pay runs, CPU, reserved memory-time and database rows, plus upstream provider usage. Memory is reserved per tool and can be lowered once for a smaller call.
Hatchet
Hatchet Cloud meters task runs, events and storage after included usage, with additional seat and enterprise terms. A spawned task is another run. Worker compute remains part of your chosen deployment.
Estimate tasks, event volume, payload storage and worker resources together. Hatchet’s orchestration price is outside the hosting calculator.
Limits that affect the choice.
- CrabTalk’s 100 ms CPU, 128 MiB memory and 60-second call-chain limits exclude large parsing jobs and long worker processes.
- Hatchet workers supply execution capacity. Task slots and concurrency limits control scheduling; they are not a fixed memory allocation per task.
- A failed external operation can be retried on either path. Keep idempotency and duplicate handling with the operation that owns the side effect.
Questions about CrabTalk and Hatchet.
Does Hatchet Cloud run all my task code?
The documented worker model uses processes you run in your chosen environment. Hatchet assigns work to registered tasks and receives their results.
Is Hatchet limited to simple queued jobs?
No. It includes durable tasks, event waits, sleeps and workflow coordination, as well as ordinary tasks and queues.
Can CrabTalk handle scheduled work?
Yes. It supports crons and scheduled wakes. Those calls remain bounded tools, with different progress and recovery behavior from a durable workflow.
Sources.
Checked .