Firebase Review 2026

What is Firebase?

Fifty thousand document reads a day. That one quota shapes a lot of Firebase upgrade decisions, and it is the first thing to understand about Firebase pricing, Spark vs Blaze plan. Spark is the no-card tier with fixed daily and monthly quotas. Blaze is pay as you go, keeps the same free allowance, and bills you only for usage above it. Google's own docs make the difference sharp: on Spark, a product that crosses its quota simply switches off until the next reset, while on Blaze it keeps running and the meter keeps ticking.

Firebase started in 2011 as a San Francisco startup for real-time sync, and Google bought it in 2014. Today it is the app-backend layer of Google Cloud: every Firebase project is also a Google Cloud project, sharing billing and IAM. Indie Android developers, React and Flutter teams, hackathon builders and agencies shipping client MVPs make up most of its users.

Firebase pricing: Spark vs Blaze plan, product by product

This table follows the official pricing page. Blaze includes every Spark allowance before charges begin.

App Hosting, built for Next.js and Angular apps with server rendering, is Blaze-only: 10 GiB of outbound data a month is free, then $0.15 to $0.20 per GiB, with Cloud Run and Cloud Build passed through at their own rates. SQL Connect (renamed from Data Connect in April 2026, with the same APIs) puts a managed Cloud SQL for PostgreSQL instance behind GraphQL. Spark projects get one three-month trial instance; after that it needs Blaze, from about $9.37 a month. Remote Config switched to usage pricing on 1 September 2026. Both plans keep 100,000 free fetches a day, and Blaze then pays $0.000006 per request up to 10 million a day and $0.000001 beyond that. Existing Blaze projects are not charged until 1 February 2027, and Spark projects that keep crossing the limit are throttled after 30 days.

The pricing page also offers $300 of free credit, if eligible, when you upgrade to Blaze. It is the Google Cloud Free Trial credit, meant for accounts new to both, and it cannot pay for Gemini Developer API usage.

Where the Firebase Spark plan free limits stop a growing app

For a to-do app or a college project, Spark is generous. The pinch points show up in a predictable order.

• No server code. Cloud Functions needs Blaze. Sending a welcome email, calling Razorpay's webhook or running a scheduled cleanup therefore forces the upgrade on its own.
• No file uploads. Firebase storage lost its Spark allowance on 3 February 2026: Google's storage FAQ says Spark projects cannot reach any Cloud Storage bucket, legacy appspot.com buckets included. A profile photo feature means Blaze.
• No phone sign-in. SMS verification is listed as Blaze-only in the Authentication limits doc.
• The 50K read wall. A feed screen that loads 40 documents, opened 1,300 times, uses the whole day. When that happens on Spark, Firestore stops answering for every app in the project until midnight Pacific time.
• 100 live connections on Realtime Database is enough for testing, not for a live quiz with a few hundred students.
• 3,000 daily users. The Auth limits page caps Spark projects that use Authentication with Identity Platform at 3,000 Tier 1 daily active users.

So, is Firebase free? For prototypes, yes, with no card at all. For a production app with uploads or backend logic, treat Blaze as the real starting point and budget for usage above the free slice.

What Firebase Authentication phone OTP cost looks like for Indian users

Firebase authentication itself is free for email, Google, Apple and anonymous sign-in up to 50,000 monthly active users. Phone login is a different matter. Identity Platform's pricing page lists SMS at $0.07 per message to India, $0.01 to the US, $0.04 to the UK and $0.45 for countries not on the list. It also says the first ten SMS each day are not billed.

The maths gets serious fast. An Indian app sending 10,000 OTPs a month, counting resends, pays close to $700 a month in SMS alone, far more than its database. Standard Firebase Authentication on Blaze also caps SMS at 3,000 a day per project, so a large consumer app has to move to Identity Platform for higher limits. Many Indian teams pair Google sign-in with phone OTP only where a verified number really matters, or route OTPs through a cheaper local SMS gateway and mint custom tokens. Turn on SMS region policy and App Check early, since bot-driven SMS pumping is a known cause of surprise charges.

Running Blaze from India: cards, rupees and regions

Can you use the Blaze plan in India? Yes. Blaze is a link to a Cloud Billing account, so the Google Cloud rules apply. Indian billing addresses are invoiced in rupees with GST by Google Cloud India Private Limited. Visa, Mastercard and Amex cards work if the bank allows recurring charges; RuPay is not on the accepted list, and UPI is limited to qualified organisations. RBI rules mean automatic card debits above ₹15,000 may be declined, so bigger bills can need a manual payment.

For latency, put Cloud Firestore in asia-south1 (Mumbai) or asia-south2 (Delhi). One catch: the free Cloud Storage quota applies only to buckets in us-central1, us-east1 and us-west1, so a Mumbai bucket is billed from the first gigabyte.

Firestore vs Realtime Database: choose before the bill arrives

The two databases charge for different things, which changes how you design screens. The realtime database bills for storage and bandwidth only, at a higher per-GB rate, and keeps everything in one large JSON tree. It handles presence (who is online) natively and suits chat typing indicators or live cursors. A single instance tops out around 200,000 concurrent connections and 1,000 writes per second, after which you shard.

Cloud Firestore bills mostly per document operation, stores data as collections of documents, supports compound queries with indexes and scales without sharding. Google recommends it for most new projects, and its offline cache covers web as well as Android and Apple clients. The practical rule: if your app reads many small documents repeatedly, Firestore costs grow with traffic; if it streams large blobs of data to many listeners, Realtime Database bandwidth costs grow instead.

Why Firebase bills spike, and the fixes that work

Most "why is my Firebase bill so high" threads trace back to a handful of mistakes:

• Snapshot listeners left open on large collections, re-reading documents on every change.
• No pagination, so each screen load fetches hundreds of documents.
• Cloud Functions that write to the same document that triggers them, creating a loop.
• Public Storage files hotlinked from other sites, driving download egress.
• Phone auth abuse from bots, as covered above.

Budget alerts in Cloud Billing only send email; they do not stop spending. Google's plan docs recommend spend caps for Cloud Functions, App Hosting, AI Logic and Extensions, which pause new usage once a budget is reached. Before deploying, run rules and functions against the Firebase Emulator Suite locally, where reads and writes cost nothing, and watch the usage tab in the console during your first week live.

Wiring Firebase Authentication into a React app

Create a project in the console, register a web app and copy the config object. Install the firebase npm package, call initializeApp with that config, then getAuth. Enable a provider such as Google under Authentication, and trigger signInWithPopup with a GoogleAuthProvider from a button. Subscribe to onAuthStateChanged in a context provider so components know who is signed in. The config's API key is an identifier rather than a secret; protection comes from Security Rules that check request.auth and from App Check.

The rest of the toolkit lives in the same console: Firebase Cloud Messaging for push notifications at no cost, Crashlytics for crash reports, Remote Config for feature flags and A/B tests, and Firebase Hosting for static sites with free SSL. Firebase AI Logic lets client apps call Gemini models through a proxy, and stays on Spark when you use the Gemini Developer API free tier. Genkit, Google's Apache 2.0 framework for AI flows in TypeScript, Go and Dart (Python in preview), covers the server side.

Firebase vs Supabase for a small product team

Which is better? Supabase wins if your data is relational, you want predictable monthly bills, or you may self-host later. Firebase wins for mobile-first apps that need push, crash reporting and offline sync without stitching services together. Other open source picks are Appwrite and PocketBase, the latter a single Go binary that suits small internal tools.

Security, compliance and the products being retired

Firebase's privacy page says all its services have completed ISO 27001 and SOC 1, 2 and 3 audits, with some also covered by ISO 27017 and 27018. Google acts as data processor under the Firebase Data Processing and Security Terms. Authentication runs only from US data centres, which matters for teams with strict data residency needs. Production safety depends mostly on you: locked-down Security Rules, App Check, and never shipping test-mode rules that allow open reads.

Google is also pruning. Firebase Studio, the browser-based AI workspace, closed new workspaces and sign-ups on 22 June 2026 and shuts down on 22 March 2027; Google points users to AI Studio or Antigravity. Test Lab was deprecated on 30 September 2026 and shuts down fully on 30 September 2027, with Google Cloud's Developer Device Platform named as the replacement. Deployed apps are unaffected, but do not build new workflows on either.

Who should use Firebase? Solo developers and small teams building Android, iOS, Flutter or web apps who value speed over schema control, and who will set spend caps before launch. Teams needing heavy SQL reporting, or a fixed monthly bill, should price Supabase first.