Nango vs. Unified.to: Build Your Own Unified API or Use One?
June 10, 2026
Nango is a platform for building your own integrations, and Unified.to is a unified API that has already built them, so the choice comes down to who writes and maintains the integration logic. Choose Nango if you want to write and own that logic in TypeScript and run it on Nango's runtime, including mappings for each customer's custom objects. Choose Unified.to if you want maintained integrations that read and write the source at request time, send change events carrying the records on supported integrations, and store no customer records, with each of its 1264 integrations returning the same normalized objects for its category.
All figures below were checked against both vendors' published documentation on 8 October 2026.
How do Nango and Unified.to compare?
| Dimension (8 Oct 2026) | Nango | Unified.to |
|---|---|---|
| What it is | Infrastructure for writing integrations: auth, proxy, a function runtime and a records cache | A unified API with normalized objects per category, maintained by Unified.to |
| Who writes the integration logic | You, as TypeScript functions, or you start from Nango's pre-built functions | Unified.to |
| Data model | You define it; one shared standard model (HRIS employees, 11 integrations) | Normalized per category and published, with typed fields and enums (e.g. Invoice: 40 fields, 10 status values, 25 line-item fields) |
| APIs covered | 1,046 in its catalogue, "1,000+ APIs across 30+ categories"; pre-built syncs or actions on 235 | 1264 across 34 categories, every one on the unified model |
| Reads | From Nango's records cache after a sync, or at request time through actions and the proxy | At request time from the source API |
| Writes | Through actions; synchronous by default, async actions queued per environment | Through the normalized model at request time |
| Customer records stored | Synced records cached on Nango Cloud (AWS, US); payloads pruned 30 days after the last update; all records deleted after 60 days without a sync run. Proxy and action logs hold no bodies | No customer records stored; logs hold operational metadata only |
| Change detection | Native webhooks forwarded where the provider has them; polling syncs down to every 30 seconds on every plan | Native webhooks where available; virtual webhooks down to one minute on paid plans, on supported integrations and objects |
| What a change event carries | Counts of added, updated and deleted records; you then fetch the records from the cache | The changed records, on supported integrations and objects |
| Database delivery | Not offered: "Do you write directly to my database? No." | Database Sync to MongoDB, MySQL, Postgres, MSSQL or MariaDB |
| Hosted regions | US only ("We currently don't have an EU cloud") | One region per workspace: US on every plan; EU or AU on Pro and Scale |
| Self-hosting | Enterprise (BYOC or self-managed); a limited free edition with auth and proxy only | Enterprise: single tenant, private cloud, dedicated cloud or on-prem |
| Source code | Public on GitHub, source-available under the Elastic License 2.0 | Platform source not published |
| MCP | Agent sessions: one MCP URL per session, US host, action functions only | One host per region (US, EU, AU), or on-prem or single tenant with the API on Enterprise; 22,566 tools |
| API rate limits | Per account, 200 to 2,000 requests a minute by plan | Per workspace, 5,000, 7,500 or 10,000 requests per rolling 60 seconds (Grow, Pro, Scale) |
| Backend SDKs | Node; Python, Java, Ruby, Go, Rust and PHP "Coming soon" | TypeScript, Python, Go, Ruby, PHP, Java, C# |
| Compliance | SOC 2 Type II, GDPR, HIPAA (BAA with the $450 Growth add-on) | SOC 2 Type II, GDPR, CCPA/CPRA, PIPEDA, HIPAA (BAA on Scale) |
What does Nango do well?
Nango is stronger when the integration itself is your product: you write the logic, keep it in your repository, and map each customer's custom objects into models you define. It also polls faster, starts free, and offers a limited self-hosted edition at no cost.
How does building on Nango work?
On Nango you write TypeScript functions that call each provider's API, deploy them to Nango's runtime, and define your own data models; Nango handles the parts around them. In its words, "Nango supports 1,000+ APIs and handles auth, credentials, execution, retries, rate limits, observability, environments, and tenant isolation" (intro to Nango). Functions live in your repository and deploy with the Nango CLI, or through a Functions API that skips the Git workflow (functions guide).
If your customers run Salesforce orgs with dozens of custom objects, or NetSuite instances with tenant-specific schema, writing the mapping yourself puts each custom object into a model you define, running on Nango's runtime. On Unified.to, custom fields come back through the Metadata API and custom objects through Passthrough, with that mapping in your own code. That is the case Nango is built for, and the engineering cost of maintaining those functions is the price of it.
What do Nango's pre-built functions cover?
Nango's docs list 8,859 pre-built functions (1,508 syncs and 7,351 actions) across 235 of its 1,046 APIs as of 8 October 2026. They are templates you can use or adapt, public on GitHub, so you can read exactly how each integration fetches data. The other 811 APIs come with authorization, the proxy and logs, and you write the logic. One standard model is shared: unified-employees maps 11 HR integrations to "the standard HRIS employee model" (BambooHR page).
How fast does Nango detect changes?
As of October 2026, Nango's polling syncs can run every 30 seconds, and its limits page says that is "available on all plans", including Free. Where a provider sends webhooks, Nango forwards them; it has routing defined for 61 providers. Unified.to's virtual webhooks run at a one-minute minimum on paid plans, so on interval alone Nango is faster.
Can you start on Nango for free, or host it yourself?
Yes to both. Nango's pricing includes a permanent Free plan, hard-capped at 10 connections, 10 compute hours and 10 GB of data transfer a month, and Pay-as-you-go starts at $50 a month, returned as $50 of usage credits. Self-hosting is on Enterprise, either inside your own AWS, GCP or Azure account (BYOC) or self-managed, and a limited free self-hosted edition covers auth and proxy only (self-hosting).
Where are Nango and Unified.to the same?
Both let you register your own OAuth app so consent screens show your brand, and both let you take your customers' tokens with you if you leave: Nango says "only your own developer app lets you export access and refresh tokens and move off Nango without users re-authorizing" (auth guide). Both handle token refresh, retries and rate limits. Neither is a reason to choose one over the other.
What does Unified.to do differently?
Unified.to has already written and maintains the integration logic, so a normalized Contact, Employee or Invoice works across every integration in its category with no function per provider. The trade is control: you work with Unified.to's model, with raw fields and Passthrough for anything it doesn't cover.
What does a maintained data model save you?
A maintained model removes the per-provider work: one request shape for every CRM, HRIS or accounting integration, with mapping, pagination and provider API changes handled by Unified.to. A team that needs 20 CRM integrations this quarter integrates once rather than writing, or adapting from Nango's templates, and maintaining 20 sets of functions. The model also covers objects Nango's templates don't: payroll is modelled as pay runs on integrations including Gusto, Paychex, Paycor and Rippling, with pay groups, pay codes and typed payslip lines for earnings, deductions, taxes and employer contributions. Nango's 8,859 pre-built functions include no pay run or payslip function.
For fields outside the model, raw returns the source's own fields on reads and accepts them on writes, the Metadata API looks up custom fields (50 platforms in the docs as of October 2026), and Passthrough reaches any provider endpoint.
Can you check coverage before you build?
Yes, for accounting: Unified.to publishes readable fields, writable fields, webhook events and list parameters for each accounting integration in its accounting integrations matrix. Nango's docs pages list each integration's pre-built syncs and actions publicly, but none is labelled read or write, and there's no field-level grid; Nango points to the dashboard's Endpoints tab for detail.
Where do customer records live?
Nango's syncs keep customer records in a cache; Unified.to stores none, because reads and writes go to the source at request time. Its privacy policy says logs "contain only operational metadata such as endpoint, timing, status, and workspace". Nango's proxy and actions also log no bodies: "Nango records the URL, method, status code, and headers — never request or response bodies" (security).
Nango's data sync page says records "sit encrypted in a cache on Nango Cloud (AWS, US) until you fetch them, are pruned 30 days after their last update, and deleted entirely if a sync stops running for 60 days", and Nango's docs describe the cache as "a delivery mechanism, not your app's long-term data store" (sync efficiency). If you sync on Nango, you also build the pipeline from its cache into your own database. Both vendors process your customers' data in transit, so both belong on your subprocessor list. The general case is in why unified APIs shouldn't store your customer data.
What do change events carry?
Unified.to's events carry the changed records; Nango's sync webhooks carry counts. Nango's webhook payload reports added, updated and deleted numbers, and you fetch the records with cursor-based reads from its cache (webhooks from Nango). On Unified.to, virtual webhooks deliver the record itself, so there's no second fetch.
If you want the records in your own database rather than in webhook handlers, Unified.to's Database Sync writes normalized data into MongoDB, MySQL, Postgres, MSSQL or MariaDB. Nango's answer to "Do you write directly to my database?" is "No" (context sync).
Where can each one host your data?
As of October 2026, Nango Cloud is US-only: its pricing page says "We currently don't have an EU cloud", and other regions need Enterprise self-hosting. Each Unified.to workspace runs in one region: the US on every plan, or the EU or AU on Pro and Scale. A product with an EU or Australian residency requirement can meet it on Unified.to's Pro or Scale plan.
Which one works better with AI agents?
Unified MCP is the better fit if your agents need normalized tools across categories from a host in your region; Nango's agent sessions fit if your agents should call only the action functions you have built or adopted, from a US host. On 6 October 2026 Nango deprecated its Connection MCP server "in favor of the new and more capable agent sessions" (changelog).
Each Nango agent session exposes its own MCP server over Streamable HTTP at https://api.nango.dev/session/{session_id}/mcp, authenticated with a session token that expires between 60 seconds and 15 days. Only action functions become tools, and agents find them with a nango_tool_search tool unless you pin them (agent sessions). Nango's MCP page cites "7,000+ ready-made tools across 1,000+ APIs".
Unified MCP runs one isolated host per region (US, EU and AU), and a credential works only on its own region's host. It exposes 22,566 tools, supports the stateless core of the 2026-07-28 MCP revision, and adds controls Nango's sessions don't document: signed connection-scoped URLs, opt-in hide_sensitive PII redaction, and Enterprise-managed authentication, where your identity provider's assertion is exchanged for a scoped token (enabled per workspace on request). Both support tool filtering and deferred tool loading.
Both vendors also ship a workspace-management MCP server (Nango's is in beta) and skills for coding agents: Nango's coding-agent skills for writing and testing its integration functions locally (as of 8 October 2026), and Unified.to's Agent Skills, with a Claude Code plugin, for building on its API.
What does each one cost?
Nango starts free and bills per connection plus usage; Unified.to starts at $750 a month and bills on API calls, with customer connections unlimited. Prices as of 8 October 2026, from Nango's pricing page and Unified.to's:
| Plan | Nango | Unified.to |
|---|---|---|
| Entry | Free: $0, 10 connections, 10 compute hours and 10 GB a month, hard-capped | Grow: from $750/mo, 750,000 API calls; all categories; unlimited connections; US region |
| Self-serve | Pay-as-you-go: $50/mo, returned as $50 of credits; $0.29 per connection, $0.72 per compute hour, $0.50 per GB | Pro: from $1,500/mo, 2M API calls; SAML SSO, secrets managers, restricted IP addresses, authorization CNAME (US and EU regions), log shipping |
| Add-on or upper tier | Growth add-on: $450/mo; SAML SSO, log export, HIPAA BAA | Scale: from $3,000/mo, 6M API calls; HIPAA BAA, 365-day logs, Multi-Region Sync |
| Top | Enterprise: custom; self-hosting, audit trail | Enterprise: priced by usage or by connection; single tenant, private cloud, dedicated cloud or on-prem |
| At low volume Nango is cheaper, and its HIPAA path ($50 plus the $450 add-on) costs less than Unified.to's Scale. The models diverge as connections grow: by our calculation from Nango's published rates, 1,000 connected customers on Nango is $290 a month in connection fees before compute and transfer, and every sync run is billed compute time even when it finds nothing. On Unified.to, connections don't add cost and a check that finds nothing isn't billed. The general case is worked through in usage-based versus per-connection pricing. |
How do you choose between Nango and Unified.to?
Three factors decide most cases: whether you want to write and own the integration logic, how many categories you need, and where your customers' data has to live.
| If you… | Choose | Why |
|---|---|---|
| Want integration logic in your own repository, in your own model | Nango | You write TypeScript functions and define the models |
| Want each customer's custom objects mapped by your own hosted code | Nango | Your functions define the model and run on Nango's runtime; on Unified.to, custom objects go through Passthrough from your code |
| Need change detection faster than one minute | Nango | Polling syncs down to 30 seconds on every plan |
| Want to start free or under $100 a month | Nango | Free plan; Pay-as-you-go from $50 |
| Need a HIPAA BAA at the lowest price | Nango | BAA with the $450 Growth add-on |
| Need to ship many integrations across categories without writing functions | Unified.to | One normalized model per category, maintained for you |
| Need payroll data (pay runs, pay codes, payslip lines) | Unified.to | Pay runs, pay codes and payslip lines in the HR model; no pre-built Nango equivalent |
| Want change events that carry the changed records | Unified.to | Nango's sync webhooks carry counts |
| Want records written into your own database | Unified.to | Database Sync; Nango doesn't write to your database |
| Have many customers with light integration use | Unified.to | Connections unlimited; Nango charges $0.29 per connection |
| Need EU or Australian hosting without Enterprise | Unified.to | EU or AU on Pro and Scale; Nango Cloud is US-only |
| Don't want a records cache in the integration layer | Unified.to | No customer records stored; Nango's syncs cache records on Nango Cloud (US), with payloads pruned 30 days after a record's last update |
| Need SDKs beyond Node | Unified.to | Seven languages; Nango's other SDKs are "Coming soon" |
| Need a hosted MCP server across categories in your region | Unified.to | Regional hosts and normalized tools; Nango's sessions run from one US host |
| Have every customer on the same one or two providers | Neither | A direct integration may be simpler than either platform |
| The two can coexist: Unified.to for the integrations its normalized model serves, Nango for the one or two where you want a customer's custom objects mapped by your own hosted code. Running two platforms has a real operational cost, so do it only where one can't cover the requirement. And if every customer uses the same provider, a direct integration may be simpler than either. For the wider build-or-buy question, see build vs buy integrations. |
Start your 30-day free trial of Unified.to or talk to our team to check the integrations your customers run.
Frequently asked questions
Is Nango a unified API?
Nango provides the infrastructure to build one: auth, a function runtime and a records cache. In its words, "In Nango, unification is optional and code-first. You define the models" (unified APIs guide). Unified.to provides the models and maintains the mapping.
Is Nango open source?
Nango describes itself as open source, and its code and templates are public on GitHub, but the licence is the Elastic License 2.0, which bars offering the software to third parties as a hosted or managed service. You can read and self-host it within those terms.
Can I move from Nango to Unified.to without customers reconnecting?
Usually, if your connections use your own OAuth app. Nango says "only your own developer app lets you export access and refresh tokens and move off Nango without users re-authorizing", and Unified.to's credential import is self-serve (how to import your integrations): you create each connection through the API with its access and refresh tokens, and activate the integration with the same OAuth app that issued them so the tokens keep refreshing. Connections made on Nango's shared OAuth apps will need your customers to authorize again. Unified.to's migration overview covers both paths, which you can mix per integration: re-authorization with your customer ID passed as external_xref, or import of credentials from your own OAuth apps.
Related reading: Architecture trade-offs: routed versus cached unified APIs · Top Merge.dev alternatives in 2026 · Paragon vs Unified.to · Usage-based versus per-connection pricing
Sources: nango.dev and nango.dev/docs, Nango's pricing page, changelog and GitHub repositories; Unified.to documentation, privacy policy and pricing page. Checked 8 October 2026.
Written by Mallory Greene.
About the author: Mallory Greene is Head of Marketing at Unified.to. She writes about integration infrastructure, unified APIs, and MCP for technical teams. Based in Toronto.