Pricing
Start free. Never pay past your plan.
Every plan comes with $2 of models a month, no API key needed, and a hard limit: nothing is billed past it.
Free
No card needed
1,000 credits / monthiA step that does work — an agent, a model call, an evaluator, JSON extraction or a tool — costs one credit; a logic step costs a tenth of one; waiting for a person costs nothing. A run costs its steps, or one credit per started minute of active time if that is more, and at least one credit. Time spent waiting for a person, or in a queue, never counts. Included credits reset monthly and do not roll over.
about 240 runs of a 6-step workflow
Build something real before you pay for anything.
- Up to 2 buildersiBuilders create and edit workflows, agents and tools; owners and admins are builders too. Viewers, who only watch and use what is built, are free, unlimited, and never count toward a plan’s builders.
- 1 workflow at a timeiHow many runs execute at the same time. Extra runs wait in line, first in, first out, and start the moment a slot frees; time in line is not charged. Up to ten times this many can wait; past that a new run is refused and a schedule skips that run and says so.
- Runs up to 30 minutesiA runaway guard, not a pricing lever: a single run stops at this limit, counted as working time — waiting for a person or in line doesn’t count. Your monthly credits and your wallet’s balance are what bound the cost.
- $2 of models a month, no API key needediEvery plan gets $2 a month of our free-tier managed models, with no API key needed. Past that, top up a prepaid balance and pay as you go for any model we offer, or bring your own provider key. Model spend never arrives as a surprise on an invoice.
- A hard limit, never a surprise billiWhen the month’s credits run out, runs pause until the next period, with warnings at 50%, 25% and 10% on the way. On a paid plan, buy more credits (they last twelve months) or move up a plan; on Free, upgrade. Nothing is ever billed past what you chose.
- Up to 3 workflows and 10 agentsiFair-use ceilings on what you can create, not what you pay for. If you genuinely need more, ask; they are raised on request.
Solo
billed monthly
5,000 credits / monthiA step that does work — an agent, a model call, an evaluator, JSON extraction or a tool — costs one credit; a logic step costs a tenth of one; waiting for a person costs nothing. A run costs its steps, or one credit per started minute of active time if that is more, and at least one credit. Time spent waiting for a person, or in a queue, never counts. Included credits reset monthly and do not roll over.
about 1,200 runs of a 6-step workflow
For one person handing off their own weekly work.
- 1 builder, +$19 each, up to 3iBuilders create and edit workflows, agents and tools; owners and admins are builders too. Viewers, who only watch and use what is built, are free, unlimited, and never count toward a plan’s builders.
- 2 workflows at a timeiHow many runs execute at the same time. Extra runs wait in line, first in, first out, and start the moment a slot frees; time in line is not charged. Up to ten times this many can wait; past that a new run is refused and a schedule skips that run and says so.
- Runs up to 1 houriA runaway guard, not a pricing lever: a single run stops at this limit, counted as working time — waiting for a person or in line doesn’t count. Your monthly credits and your wallet’s balance are what bound the cost.
- 1 website chat widgetiAn agent on your website as a chat widget, in your colors, with your logo, answering from the knowledge you give it.
- $2 of models a month, then pay as you goiEvery plan gets $2 a month of our free-tier managed models, with no API key needed. Past that, top up a prepaid balance and pay as you go for any model we offer, or bring your own provider key. Model spend never arrives as a surprise on an invoice.
- A hard limit, never a surprise billiWhen the month’s credits run out, runs pause until the next period, with warnings at 50%, 25% and 10% on the way. On a paid plan, buy more credits (they last twelve months) or move up a plan; on Free, upgrade. Nothing is ever billed past what you chose.
- Up to 10 workflows and 25 agentsiFair-use ceilings on what you can create, not what you pay for. If you genuinely need more, ask; they are raised on request.
Team
billed monthly
25,000 credits / monthiA step that does work — an agent, a model call, an evaluator, JSON extraction or a tool — costs one credit; a logic step costs a tenth of one; waiting for a person costs nothing. A run costs its steps, or one credit per started minute of active time if that is more, and at least one credit. Time spent waiting for a person, or in a queue, never counts. Included credits reset monthly and do not roll over.
about 6,000 runs of a 6-step workflow
For a team running workflows other people depend on.
- 5 builders, +$19 each, up to 25iBuilders create and edit workflows, agents and tools; owners and admins are builders too. Viewers, who only watch and use what is built, are free, unlimited, and never count toward a plan’s builders.
- 8 workflows at a timeiHow many runs execute at the same time. Extra runs wait in line, first in, first out, and start the moment a slot frees; time in line is not charged. Up to ten times this many can wait; past that a new run is refused and a schedule skips that run and says so.
- Runs up to 2 hoursiA runaway guard, not a pricing lever: a single run stops at this limit, counted as working time — waiting for a person or in line doesn’t count. Your monthly credits and your wallet’s balance are what bound the cost.
- 5 website chat widgetsiAn agent on your website as a chat widget, in your colors, with your logo, answering from the knowledge you give it.
- Sandboxed Python stepsiRun your own Python inside a workflow, isolated from the engine. Team, Growth and Business.
- $2 of models a month, then pay as you goiEvery plan gets $2 a month of our free-tier managed models, with no API key needed. Past that, top up a prepaid balance and pay as you go for any model we offer, or bring your own provider key. Model spend never arrives as a surprise on an invoice.
- A hard limit, never a surprise billiWhen the month’s credits run out, runs pause until the next period, with warnings at 50%, 25% and 10% on the way. On a paid plan, buy more credits (they last twelve months) or move up a plan; on Free, upgrade. Nothing is ever billed past what you chose.
- Up to 100 workflows and 200 agentsiFair-use ceilings on what you can create, not what you pay for. If you genuinely need more, ask; they are raised on request.
Growth
billed monthly
50,000 credits / monthiA step that does work — an agent, a model call, an evaluator, JSON extraction or a tool — costs one credit; a logic step costs a tenth of one; waiting for a person costs nothing. A run costs its steps, or one credit per started minute of active time if that is more, and at least one credit. Time spent waiting for a person, or in a queue, never counts. Included credits reset monthly and do not roll over.
about 12,100 runs of a 6-step workflow
For a business putting AI to work across several teams.
- 10 builders, +$19 each, up to 50iBuilders create and edit workflows, agents and tools; owners and admins are builders too. Viewers, who only watch and use what is built, are free, unlimited, and never count toward a plan’s builders.
- 12 workflows at a timeiHow many runs execute at the same time. Extra runs wait in line, first in, first out, and start the moment a slot frees; time in line is not charged. Up to ten times this many can wait; past that a new run is refused and a schedule skips that run and says so.
- Runs up to 4 hoursiA runaway guard, not a pricing lever: a single run stops at this limit, counted as working time — waiting for a person or in line doesn’t count. Your monthly credits and your wallet’s balance are what bound the cost.
- 10 website chat widgetsiAn agent on your website as a chat widget, in your colors, with your logo, answering from the knowledge you give it.
- Sandboxed Python stepsiRun your own Python inside a workflow, isolated from the engine. Team, Growth and Business.
- $2 of models a month, then pay as you goiEvery plan gets $2 a month of our free-tier managed models, with no API key needed. Past that, top up a prepaid balance and pay as you go for any model we offer, or bring your own provider key. Model spend never arrives as a surprise on an invoice.
- A hard limit, never a surprise billiWhen the month’s credits run out, runs pause until the next period, with warnings at 50%, 25% and 10% on the way. On a paid plan, buy more credits (they last twelve months) or move up a plan; on Free, upgrade. Nothing is ever billed past what you chose.
- Up to 250 workflows and 500 agentsiFair-use ceilings on what you can create, not what you pay for. If you genuinely need more, ask; they are raised on request.
Business
billed monthly
150,000 credits / monthiA step that does work — an agent, a model call, an evaluator, JSON extraction or a tool — costs one credit; a logic step costs a tenth of one; waiting for a person costs nothing. A run costs its steps, or one credit per started minute of active time if that is more, and at least one credit. Time spent waiting for a person, or in a queue, never counts. Included credits reset monthly and do not roll over.
about 36,500 runs of a 6-step workflow
For workflows that are part of how the company runs.
- 15 builders, +$25 each, up to 100iBuilders create and edit workflows, agents and tools; owners and admins are builders too. Viewers, who only watch and use what is built, are free, unlimited, and never count toward a plan’s builders.
- 25 workflows at a timeiHow many runs execute at the same time. Extra runs wait in line, first in, first out, and start the moment a slot frees; time in line is not charged. Up to ten times this many can wait; past that a new run is refused and a schedule skips that run and says so.
- Runs up to 6 hoursiA runaway guard, not a pricing lever: a single run stops at this limit, counted as working time — waiting for a person or in line doesn’t count. Your monthly credits and your wallet’s balance are what bound the cost.
- 25 website chat widgetsiAn agent on your website as a chat widget, in your colors, with your logo, answering from the knowledge you give it.
- Audit log exportiOwners and admins download the workspace’s audit trail as CSV or JSON, for any range of days, to keep for their own records.
- Sandboxed Python stepsiEverything Growth has, including your own Python inside a workflow, isolated from the engine.
- $2 of models a month, then pay as you goiEvery plan gets $2 a month of our free-tier managed models, with no API key needed. Past that, top up a prepaid balance and pay as you go for any model we offer, or bring your own provider key. Model spend never arrives as a surprise on an invoice.
- A hard limit, never a surprise billiWhen the month’s credits run out, runs pause until the next period, with warnings at 50%, 25% and 10% on the way. On a paid plan, buy more credits (they last twelve months) or move up a plan; on Free, upgrade. Nothing is ever billed past what you chose.
- Single sign-on, coming sooniSigning in through your identity provider is being built. It is not available yet, on any plan.
- Up to 500 workflows and 1,000 agentsiFair-use ceilings on what you can create, not what you pay for. If you genuinely need more, ask; they are raised on request.
Custom
Priced to your volume
Committed credit volume
For teams putting SynapseOrch on a critical path.
- Committed credits at a negotiated rate
- Dedicated concurrency
- Named run limits
- A contract that says so
We’ll build your first workflow with you, free.
Bring the job that already runs every week. We set it up with you in your own workspace, at no charge, with the scope agreed in writing first. In return we ask for real use and blunt feedback. Open to the first 10 businesses.
How credits work
You pay for what runs. Not for what you draw.
Workflows, agents, tools and connections cost nothing to create, and viewers are free on every plan. What runs is metered in credits, Build with AI's chat included. Every plan includes managed models, so you do not need an API key from anyone to get started.
| Agent, LLM, Evaluator, Extract JSON, Tool | 1 credit |
| Transform, If / Else, Switch, Loop, Parallel, Merge, Print, End, Retrieve, Ingest | 0.1 credits |
| Human | 0 credits |
Viewer seats are free on every plan. A step that does work — an agent, a model call, an evaluator, JSON extraction or a tool — costs one credit; a logic step costs a tenth of one; waiting for a person costs nothing. A run costs its steps, or one credit per started minute of active time if that is more, and at least one credit. Time spent waiting for a person, or in a queue, never counts. Included credits reset monthly and do not roll over. Managed model usage draws on a prepaid wallet, billed separately from plan credits, so model spend is always something you topped up rather than something that surprises you.
- GatherNo model$0
- ScorecardNova Lite$0.0006
- Open issuesTitan Embed v2$0.0001
- Write reviewClaude Sonnet 4.5$0.0104
- ApproveA person$0
- Send to opsNo model$0
Self-hosting
Or run it yourself, for nothing
SynapseOrch is AGPL-3.0. Self-hosting is not a trial and not a tier: it is the same engine, with no licence check and no telemetry.
- What it costs
- Nothing to us. You pay for your own infrastructure and your own model usage, and that is the whole bill.
- What you run
- The orchestration engine, the tool servers and the UI. Postgres and Redis if you want scale mode; a JSON file store is enough to start.
- What you give up
- We do not operate it, so uptime, upgrades, backups and scaling are yours. There are no workspaces or roles in a single install.
Questions
The things people ask first
- What is a credit?
- A step that does work — an agent, a model call, an evaluator, JSON extraction or a tool — costs one credit; a logic step costs a tenth of one; waiting for a person costs nothing. A run costs its steps, or one credit per started minute of active time if that is more, and at least one credit. Time spent waiting for a person, or in a queue, never counts. Included credits reset monthly and do not roll over. Creating agents, workflows, tools or connections consumes nothing; Build with AI's chat and every run are metered: you are billed for what runs, not for what you draw.
- Why not charge per seat?
- Because seats do not map to what a workflow costs us to run. A team of three running a thousand workflows a day costs far more to serve than a team of twenty running one. Builder seats are still priced, because a builder is a real cost to support, but viewers are free and always will be.
- What happens when I run out of credits?
- Runs pause until the next period, on every plan, with warnings at 50%, 25% and 10% on the way. On a paid plan, buy more credits, which last twelve months, or move up a plan; on Free, upgrade. Nothing is ever billed past what you chose: a hard stop rather than a surprise bill.
- Do I need an API key from a model provider?
- No. Every plan can call models we operate, and every plan gets $2 of our free-tier models each month. Past that, model use draws on a prepaid balance you top up, separate from your plan credits, so model spend is always something you chose rather than something that arrived on an invoice. Or bring your own provider key and pay the provider directly.
- Can I bring my own API keys instead?
- Yes. Store your own provider credentials and workflows will use those. They are encrypted individually and decrypted only to make the call you asked for.
- Is self-hosting crippled?
- No. It is the same engine under AGPL-3.0, with no licence check, no telemetry and no feature gate. What Cloud adds is that we run it, plus the things that only make sense for a hosted multi-tenant service: workspaces, roles, an audit trail and managed models.
Something not answered here? Email info@synapseorch.com.