Runtimes & workflows / Shortlist
Temporal Cloud alternatives
Start with what you need to change about Temporal. This shortlist separates comparable tools from complementary ones, with decision criteria and official references.
Before you switch
Execution semantics
Map retries, timeouts, concurrency, scheduling, durable waits, and idempotency to the alternative. A similar API name does not guarantee equivalent behavior.
Migration boundary
Identify who runs the broker, scheduler, and worker compute. Test how in-flight work, event history, and external side effects will survive a cutover.
Actual bill
Model runs, steps, retries, compute duration, storage, and support. Reconcile included usage and limits before comparing monthly totals.
Keeping BullMQ is a separate decision
Your shortlist
Other runtimes and workflow platforms to evaluate
01 / Runtimes & workflows
Inngest
Choose Inngest when durable steps, sleeps, event-driven waits, and managed orchestration justify adopting its SDK. Existing BullMQ teams can also adopt it for workflows whose requirements exceed their current queue design.
Pricing model. As of 2026-10-20: Free $0 with 50k executions; execution pauses at the quota. Pro from $99/mo includes 1M executions. Business from $499/mo includes 10M. An execution is one function run or one step (
Executions = runs × (steps + 1)). Pro overage is tiered from $0.000050 down to $0.000015 per execution; tier breakpoints are not published. Sandbox compute is extra.02 / Runtimes & workflows
Trigger.dev
Choose Trigger.dev when you want managed compute for long TypeScript tasks, no worker fleet, and you accept deploying task code onto their machines.
Pricing model. As of 2026-10-20: Free $0 with $5 usage credit. Hobby $10/mo with $10 credit. Pro $50/mo with $50 credit. Run fee $0.000025. Default small-1x machine $0.0000338/sec. Official 10-second example: $0.000363 per run. Pro concurrency add-on: $10/mo per 50 runs above 200. HIPAA BAA add-on $500/mo on Hobby and Pro.
03 / Runtimes & workflows
Hatchet
Choose Hatchet when you want durable task orchestration with workers you control, Cloud or MIT self-hosting, and published per-task pricing that excludes retries. A Hatchet run and a Temporal Action are different billing units.
Pricing model. As of 2026-10-20, with a card on file: 1M task runs and 1M events free every month, then $30 per million runs and $5 per million events. Failed runs count; retries do not. Without a card: 2,000 runs and 2,000 events per day. Custom plans for 100M+ runs offer volume discounts that are not published.
04 / Runtimes & workflows
Upstash QStash
Choose QStash for serverless or edge HTTP handlers, scheduled calls, and fan-out to URLs without running Redis.
Pricing model. As of 2026-10-20: pay-as-you-go $1 per 100K messages. Free: 1,000 messages/day. Fixed plans: 1M messages/day for $180/mo, 10M/day for $420/mo. The pricing FAQ says each delivery attempt, including each retry, is billed as a message; the pricing markdown also says retries are free. Confirm with Upstash before budgeting retry-heavy workloads.
05 / Runtimes & workflows
Amazon SQS + Step Functions
Choose SQS when you are already on AWS, want a managed broker with an AWS SLA, and will write AWS-native workers.
Pricing model. As of 2026-10-20 (AWS Price List published 2026-09-11): SQS Standard in us-east-1 is $0.40 per million requests after 1M free requests per month; FIFO is $0.50 per million. Step Functions Standard is $0.000025 per state transition after 4,000 free per month. Express is $1 per million requests plus duration.
06 / Runtimes & workflows
Google Cloud Tasks
Choose Cloud Tasks when you are on Google Cloud and want a managed service to push HTTP tasks to your handlers.
Pricing model. As of 2026-10-20: first 1M billable operations per month free, then $0.40 per million. A billable operation is an API call or push delivery attempt, chunked at 32 KB. Regular network charges apply to HTTP targets.
07 / Runtimes & workflows
Cloudflare Queues
Choose Cloudflare Queues when your application runs on Workers and you want messages to stay on Cloudflare.
Pricing model. As of 2026-10-20: Workers Paid ($5/mo minimum) includes 1M queue operations per month, then $0.40 per million. Each write, read, and delete counts per 64 KB; typically 3 operations per message, and each retry adds a read. Workers Free includes 10,000 operations per day.
08 / Runtimes & workflows
Graphile Worker
Choose Graphile Worker when you want jobs in PostgreSQL next to your primary data, not in Redis.
Pricing model. As of 2026-10-20: open source (MIT), $0. Worker Pro $100/mo, or $25+/mo via GitHub Sponsors during the early-access launch offer.
Common questions
How should I choose a Temporal alternative?
Start with the capability you need to replace. Compare products within the same runtime or observability category, then evaluate migration effort, failure behavior, billing units, and the operating work you retain.
When should I keep Temporal?
Choose Temporal when you think in Workflows and Activities, need timers, signals, and queries, and will run Workers against a dedicated orchestration service.
Is one Temporal Action one job?
No. Starting a Workflow, starting or retrying an Activity, Heartbeats that reach the server, Timers, Signals, and Queries each count. Each Schedule execution accrues three Actions. Compare units before comparing dollars.
Sources & methodology
Connect Redis. Queues appear.
Inspect your Redis-backed BullMQ queues in one place. Hosted Durabull is free to start.