Workflow orchestration / Hatchet alternative

Hatchet.run alternative

Hatchet orchestrates tasks and durable workflows, recording completed steps so work can recover after application or worker failures. Its workers are processes you run. That workflow durability is useful for long-running agents, while their installed Linux tools and filesystem state still need an execution/storage layer.

Start buildingCompare capabilities ↓

Reviewed 2026-10-18 · Published by the Sandbox0 team

When Hatchet fits

Hatchet fits teams needing scheduling, retries, event waits, task routing, workflow visibility and durable application checkpoints. Keep it when those orchestration semantics are required; a saved Linux machine alone does not implement the same workflow engine.

When to consider Sandbox0

Consider Sandbox0 for the worker/agent environment with durable whole RootFS, online growth and supported paused-file access. Retain or implement task scheduling, checkpoints, retry/idempotency handling and workflow observability separately.

Capabilities

Hatchet and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeHatchetLong-running durable workflows are supported; worker-machine lifetime and state are separate policies.View evidence ↗Sandbox0ttl and hard_ttl default to 0 (disabled). Configure idle pause and deletion policies for the same persistent sandbox identity; an idle running sandbox still accrues compute charges.Read documentation →
Persistent filesHatchetDurable workflow records are explicit; worker-host disk retention is separately owned.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSHatchetTask/event checkpoints are not a managed complete writable Linux RootFS contract.View evidence ↗Sandbox0Persistence extends beyond /workspace to paths such as /usr/local, /etc, and /home. Runtime tmpfs and virtual filesystems are excluded.Read documentation →
Release compute, keep stateHatchetWorkflow waiting/recovery is distinct from arbitrary machine/process checkpoint-resume.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateHatchetCompleted-step replay does not establish arbitrary RAM capture.View evidence ↗Sandbox0Experimental memory checkpoints can preserve compatible process state. Use memory: true on both pause and resume on a compatible deployment. A filesystem checkpoint alone does not preserve running processes.Read documentation →

Online RootFS growth

On growth-enabled images, Sandbox0 automatically expands the writable RootFS as free space runs low, within a configured sparse logical ceiling up to 64 TiB. Agent processes keep running. Growth is enabled in the qualified Cloud region; confirm activation for other regions. Existing images and snapshots retain their captured storage geometry. The ceiling is logical capacity, not preallocated host disk.

Hatchet’s durable workflow recovery is acknowledged. It retains application checkpoints while worker compute and disks follow your chosen backend. Sandbox0 can provide durable growing RootFS for those workers; an event log and a machine filesystem solve different state-retention needs.

Files while compute is paused

Sandbox0's Files API reads persistent files without resuming compute. Filesystem-only pauses also permit edits; retained-memory pauses are read-only and reject mutations with 409. SSH, SFTP, file watching, and runtime-only mounts require running compute.

Workflow history and queued state remain available through the control plane independently of a particular worker. That is useful application data, not an interface to every path of a paused worker RootFS. Sandbox0 supplies supported paused file operations at the runtime layer.

See paused file operations →

Durable execution retains application progress

The engine records task invocations and completed child steps in a durable event log. On recovery it can resume workflow progress without rerunning completed children. This is a meaningful application-level guarantee; it does not checkpoint every worker process’s RAM or complete installed Linux filesystem.

Source 1 ↗

Workers and worker slots do not describe allocated VM resources

The pricing table defines workers as processes you run that connect to Hatchet, and slots as concurrent task capacity across them. A 100-worker allowance is not 100 hosted Linux machines or a CPU/RAM allocation. Your execution hosts and model services remain part of the system budget.

Source 1 ↗

Count task runs, events and retries under their actual rules

One task execution is one run; spawning a child task counts another run. A failed run still counts, but the FAQ says retries do not add run charges. Events that trigger workflows or satisfy waits have a separate meter. The September 29, 2026 announcement retires older self-serve plan names in favor of pay-as-you-go.

Source 1 ↗Source 2 ↗

A card changes allowances and capacity limits

The detailed table lists 2,000 runs and 2,000 events per day without a card, with 10 workers and one-day history. With a card it lists one million runs and events per month, 100 workers and seven-day history. The headline free-million offer should not be applied to every no-card account.

Source 1 ↗

Pricing

Compare billing models, then price your workload

CPU allocation, memory, billing minimums, disk, plan fees, and supported workload differ. These meters are not equal-performance machine quotes. Confirm region, resource units, and included allowances before estimating a total.

Hatchet

Pricing reviewed 2026-10-18

Task runs with card
$30 per million beyond 1M/month
Failed runs count; retries do not add runs under the FAQ.
Rate source ↗
Events with card
$5 per million beyond 1M/month
A separate allowance/meter from task executions.
Rate source ↗
Active payload storage
$0.03 per GB-hour above included storage
Card-backed tier includes 10 GB; inputs/outputs are held for 24 hours.
Rate source ↗
Logs and traces overage
$0.0004 per GB-hour as labeled
Table lists 10 GB included and describes ingestion volume; confirm the effective retention/meter.
Rate source ↗
No-card allowance
2,000 runs and 2,000 events per day
Lower worker, history, storage and workspace limits than card-backed PAYG.
Rate source ↗

Card-backed PAYG includes 50 users, then the page lists $10 per additional user. Payload storage, logs/traces and seats are separate from runs/events and from your worker compute. The pricing table mixes a monthly ingestion description with a GB-hour label for log overage, so confirm that meter before estimating retention-heavy use. The MIT-licensed self-hosted engine still requires infrastructure and operations.

Check current Hatchet pricing ↗

Sandbox0

$0.015/GiB-hour compute (billed by the second at approximately $0.000004167/GiB-second), plus $0.02/GiB-month of retained incremental RootFS and memory checkpoint data. Shared base images and platform recovery retention are excluded; paused sandboxes accrue storage charges only.

Compute / GiB-hour
$0.015
Retained data / GiB-month
$0.02
Paused compute
$0.00

A 4 GiB agent, 100 running hours

$6.00 compute

4 GiB × 100 hours, billed by the second. Keeping 20 GiB of billable incremental data for a full billing month adds $0.40. CPU is derived from memory by the platform; choose a tier that fits your workload.

Understand storage billing →

Before migrating from Hatchet

  1. Inventory tasks/child tasks, events, durable waits, retry semantics, worker ownership, card status and history/storage requirements.
  2. Retain the workflow engine as needed and integrate compatible Sandbox0 workers with explicit lifecycle and idempotent task boundaries.
  3. Test workflow recovery alongside system-package retention, RootFS growth, paused files and combined run/event/storage/worker/model costs.

Common questions

Is Sandbox0 a drop-in Hatchet replacement?

Evaluate API compatibility and product scope. These products serve different core workflows, so a change may require a different architecture.

Does keeping files also keep a running agent?

Filesystem state and volatile process state are separate. Sandbox0 requires explicit compatible memory checkpoints to preserve running processes; its default filesystem resume starts fresh processes.

Sources and review scope

The Boat team's Battleships directory is the discovery source. References below describe the product and plan discussed on this page. Check current availability, regional terms and billing units before migrating. Where the cited material does not establish a capability or price, the comparison identifies that uncertainty.

Battleships research record ↗hatchet.run/platform/durable-execution ↗hatchet.run/pricing ↗hatchet.run/announcement/pay-as-you-go-pricing ↗Sandbox0 RootFS capacity →

Related comparisons

Expanse Cluster Intelligence vs Sandbox0 →Google Agent Runtime vs Sandbox0 →Hyrex vs Sandbox0 →Mastra vs Sandbox0 →Microsoft Foundry hosted agents vs Sandbox0 →Nextmv DecisionOps vs Sandbox0 →