Side-by-side comparison of 10+ serverless free tiers in 2026. Compare AWS Lambda, Google Cloud Functions, Azure Functions, Cloudflare Workers, Deno Deploy, Cloud Run, Val Town, and more — invocations, compute, cold starts, and billing models.

Serverless Free Tier Comparison 2026

Published 2026-03-31 · Not yet reviewed · Data verified from our index of 1,547 developer tools · 10+ serverless platforms compared

10+
Serverless Platforms
Workers
Best Edge / Low Latency
Lambda
Most Invocations Free
Cloud Run
Most Generous Overall

Quick verdict: AWS Lambda leads on raw invocation count (1M/month free) and has the largest ecosystem. Google Cloud Run is the most generous overall with 2M requests, 360K GB-seconds, and 1 GB free egress. Cloudflare Workers dominates edge computing with 100K requests/day, sub-5ms cold starts, and the unique advantage of CPU-time billing (you don’t pay for I/O waits). Azure Functions matches Lambda at 1M requests/month with a consumption-based model.

The critical distinction most developers miss: Cloudflare Workers bills by CPU time while AWS Lambda, Google Cloud Functions, and Azure Functions bill by wall-clock time (including time spent waiting for network I/O, database queries, and external API calls). For I/O-heavy workloads, this can mean 10-50x cheaper effective compute on Workers. This single billing model difference matters more than headline invocation counts.

Jump to section

  1. Main Comparison Table
  2. Traditional FaaS (Lambda, Cloud Functions, Azure Functions)
  3. Edge Compute (Workers, Vercel Edge, Deno Deploy)
  4. Full-Service Serverless (Cloud Run, App Runner)
  5. Specialized Serverless (Val Town, DBOS, Inngest, Trigger.dev)
  6. The Billing Gotcha: CPU-Time vs Wall-Clock-Time
  7. Best for Each Use Case
  8. Hidden Costs and Gotchas
  9. Pricing Change Timeline
  10. Data Source

Main Comparison Table

Side-by-side comparison of the top serverless free tiers.

Platform Free Invocations Free Compute Memory / CPU Timeout Billing Model Cold Start Always Free?
AWS Lambda MOST INVOCATIONS CHANGED JUN 30 Source ↗ 1M / month 400K GB-sec/mo 128 MB – 10 GB 15 min max Wall-clock time 100–500ms (cold) Yes (Always Free)
Google Cloud Functions Source ↗ 2M / month 400K GB-sec + 200K GHz-sec 128 MB – 32 GB 9 min (1st gen) / 60 min (2nd gen) Wall-clock time 200–800ms (cold) Yes (Always Free)
Azure Functions Source ↗ 1M / month 400K GB-sec/mo 1.5 GB max (Consumption) 5 min (default) / 10 min (max) Wall-clock time 200–1000ms (cold) Yes (Always Free)
Google Cloud Run MOST GENEROUS Source ↗ 2M / month 360K GB-sec + 180K vCPU-sec Up to 32 GB / 8 vCPU 60 min max Wall-clock time (request-based) 100–500ms (cold) Yes (Always Free)
Cloudflare Workers BEST EDGE Source ↗ 100K / day (~3M/mo) 10ms CPU time / invocation 128 MB RAM 30s (HTTP) / 15 min (Cron) CPU time only <5ms (near-zero) Yes
Vercel Edge Functions CHANGED SEP 7 Source ↗ 1M edge requests/mo 4 hrs Active CPU/mo 128 MB (Edge) / 1 GB (Serverless) 25s (Hobby) Wall-clock time (credit-based) <5ms (Edge) / 250ms (Serverless) Yes (Hobby)
Deno Deploy CHANGED AUG 28 Source ↗ 1M / month 10 hrs CPU time/mo 512 MB RAM No hard limit CPU time <10ms (near-zero) Yes
Val Town Source ↗ 100K runs/day 1 min wall clock / run Managed (V8 isolates) 1 min per run Run-based <50ms Yes
DBOS CHANGED SEP 3 Source ↗ 1M service calls/mo 1 microVM (512 MB, 1 vCPU) 512 MB / 1 vCPU No hard limit Service call-based Scale-to-zero (seconds) Yes
Inngest CHANGED AUG 28 Source ↗ 50K executions/mo 100K events/mo Runs on your serverless Configurable retries Execution-based Depends on host Yes (Hobby)
Trigger.dev Source ↗ $5 free compute/mo 20 concurrent runs Managed compute No hard limit Compute time-based Warm pool available Yes
Key takeaway: The main comparison table reveals two distinct models. Traditional FaaS (Lambda, Cloud Functions, Azure Functions) bills by wall-clock time with generous monthly invocation counts. Edge platforms (Workers, Deno Deploy) bill by CPU time with near-zero cold starts. Cloud Run bridges the gap as a container-based serverless with the most generous free tier overall. For background jobs, Inngest and Trigger.dev offer serverless workflow orchestration rather than raw compute.

Traditional FaaS

The big three cloud providers’ serverless compute offerings. These are the most mature platforms with the largest ecosystems, but they all bill by wall-clock time.

Provider Free Invocations Free Compute Ecosystem Languages Best For
AWS Lambda LARGEST ECOSYSTEM CHANGED JUN 30 Source ↗ 1M / month 400K GB-sec API Gateway, DynamoDB, S3, SQS, SNS, EventBridge Node.js, Python, Go, Java, .NET, Rust, Ruby AWS-native backends, event-driven architectures
Google Cloud Functions MOST FREE INVOCATIONS Source ↗ 2M / month 400K GB-sec + 200K GHz-sec Cloud Run, Firestore, Pub/Sub, Cloud Storage Node.js, Python, Go, Java, .NET, Ruby, PHP GCP-native, Firebase integrations
Azure Functions Source ↗ 1M / month 400K GB-sec Cosmos DB, Service Bus, Event Grid, Storage Node.js, Python, Java, .NET, PowerShell Microsoft/.NET shops, enterprise integrations
AWS Lambda vs Google Cloud Functions vs Azure Functions: All three offer similar free compute (400K GB-seconds). Google Cloud Functions leads on invocations (2M vs 1M). AWS Lambda has the largest ecosystem and most third-party integrations. Azure Functions is the natural choice for .NET/Microsoft shops. Important: All three charge for API Gateway separately — Lambda invocations are free, but the API Gateway to expose them adds $3.50/million requests (AWS). Google Cloud Functions includes HTTP triggers for free; Azure Functions uses Azure Functions Premium for custom domains.

Edge Compute

Serverless platforms that run code at the edge (CDN PoPs worldwide). Near-zero cold starts, global distribution, but more constrained runtimes.

Provider Free Tier Cold Start Runtime Storage Best For
Cloudflare Workers BEST EDGE Source ↗ 100K req/day, 10ms CPU/req <5ms V8 isolates (JS/TS/Wasm) KV: 1 GB, D1: 5 GB, R2: 10 GB APIs, edge logic, full-stack apps
Vercel Edge Functions CHANGED SEP 7 Source ↗ 1M edge req/mo, 4 hrs CPU <5ms V8 isolates (JS/TS) Blob: 1 GB, KV: via Vercel KV Next.js middleware, edge API routes
Deno Deploy CHANGED AUG 28 Source ↗ 1M req/mo, 10 hrs CPU, 20 GiB egress <10ms Deno runtime (JS/TS, Web APIs) KV: 1 GB (450K reads/mo) Deno/Fresh apps, Web API-native code
Cloudflare Workers leads the edge: The combination of 100K daily requests, near-zero cold starts, and integrated storage (KV, D1, R2, Durable Objects, Queues) makes Workers the most complete edge platform. Vercel Edge Functions are ideal if you’re already on Vercel/Next.js. Deno Deploy offers the most developer-friendly experience with native TypeScript, Web APIs, and built-in KV storage. All three use V8 isolates for instant startup — a fundamentally different architecture from container-based FaaS.

Full-Service Serverless

Container-based serverless that runs any Docker image — more flexibility than FaaS, with scale-to-zero on the free tier.

Provider Free Tier Container Support Concurrency Egress Best For
Google Cloud Run BEST CONTAINER SERVERLESS Source ↗ 2M req/mo, 360K GB-sec, 180K vCPU-sec Any Docker image Up to 1000 concurrent per instance 1 GB North America free Containerized APIs, microservices
AWS App Runner CHANGED JUN 30 Source ↗ No always-free tier (trial credits only) Any Docker image or source code Configurable Standard AWS egress Simple container deployments on AWS
Cloud Run is the clear winner here — it’s the only container-based serverless with a genuinely generous always-free tier. You can run any Docker image, handle up to 1000 concurrent requests per instance, and the 2M requests/month free tier is competitive with any FaaS. AWS App Runner doesn’t have an always-free tier, making it a non-starter for free-tier comparisons. For containerized workloads that need more than FaaS constraints allow, Cloud Run is the answer.

Specialized Serverless

Purpose-built serverless platforms for specific workloads — scripting, durable workflows, background jobs, and orchestration.

Provider Free Tier Specialty Durability Best For
Val Town Source ↗ 100K runs/day, unlimited public vals Script-as-a-service, shareable functions SQLite storage, blob storage Quick scripts, prototypes, public APIs, MCP servers
DBOS CHANGED SEP 3 Source ↗ 1M service calls/mo, 1 microVM Durable workflow orchestration Built-in (transactional, exactly-once) Transactional workflows, reliable background processing
Inngest CHANGED AUG 28 Source ↗ 50K executions/mo, 100K events Event-driven workflow orchestration Built-in retries, step functions Background jobs, scheduled tasks, event-driven workflows
Trigger.dev Source ↗ $5 free compute/mo, 20 concurrent Background jobs with managed compute Durable execution, retries Long-running background jobs, scheduled tasks
Cloudflare Durable Objects CHANGED SEP 2 Source ↗ 100K requests/day (on Workers free) Stateful serverless compute Strongly consistent, WebSocket hibernation Real-time collaboration, stateful edge computing
Val Town is the easiest way to get serverless functions running — write a function, get a URL, share it with anyone. Great for prototyping and small tools. DBOS solves a hard problem: durable, exactly-once execution with transactional guarantees. Inngest and Trigger.dev both handle background jobs but with different models — Inngest is event-driven (runs on your serverless), Trigger.dev provides managed compute. Cloudflare Durable Objects uniquely enables stateful serverless at the edge.

The Billing Gotcha: CPU-Time vs Wall-Clock-Time

This is the most important and least understood difference between serverless platforms. It can change your effective cost by 10-50x.

Wall-clock-time billing (AWS Lambda, Cloud Functions, Azure Functions, Cloud Run)

You pay for the total elapsed time of your function, including time spent waiting for database queries, HTTP requests, file I/O, and external API responses. A function that takes 500ms but spends 480ms waiting for a database query still costs 500ms of compute.

CPU-time billing (Cloudflare Workers, Deno Deploy)

You pay only for actual CPU cycles used. Time spent waiting for I/O is free. That same function that takes 500ms but only uses 20ms of CPU time costs only 20ms. For I/O-heavy workloads (most web APIs), this is dramatically cheaper.

Concrete example: An API endpoint that queries a database (200ms), calls an external API (150ms), and does 10ms of processing:

AWS Lambda: Billed for ~360ms wall-clock time at 128 MB = 0.000046 GB-seconds
Cloudflare Workers: Billed for ~10ms CPU time only

At scale (1M requests/month), this difference is significant. The Lambda function uses ~46 GB-seconds per million requests. The Workers function uses ~10,000ms (10 seconds) of total CPU per million requests — well within the free tier’s 10ms-per-request limit. For I/O-heavy APIs, Workers can be effectively 10-50x cheaper than Lambda.

Best for Each Use Case

When to Pick Each Serverless Platform

API backends (high traffic) → AWS Lambda or Google Cloud Run

Lambda for event-driven microservices in the AWS ecosystem. Cloud Run for containerized APIs with the most generous free tier (2M req/month). Both handle production-scale traffic. Cloud comparison →

Edge / low-latency APIs → Cloudflare Workers

100K requests/day with sub-5ms cold starts, globally distributed. CPU-time billing makes I/O-heavy workloads dramatically cheaper. Integrated KV, D1, R2, and Durable Objects storage. View Workers details →

Quick scripts and prototypes → Val Town or Deno Deploy

Val Town: write a function, get a URL. 100K runs/day free with MCP integration. Deno Deploy: full TypeScript runtime with Web APIs, 1M requests/month. Both offer instant deployment with no build step.

Next.js / Vercel projects → Vercel Edge Functions

Seamless integration with Next.js middleware and edge API routes. 1M edge requests/month on the Hobby plan. Use Vercel’s serverless functions for Node.js APIs, edge functions for latency-sensitive routes.

Existing cloud customers → Match your IaaS

On AWS? Use Lambda. On GCP? Use Cloud Functions or Cloud Run. On Azure? Use Azure Functions. The free tiers are similar enough that ecosystem integration and team expertise matter more than marginal differences in invocation limits.

Background jobs and workflows → Inngest or Trigger.dev

Inngest: event-driven workflow orchestration with 50K executions/month free. Trigger.dev: managed compute for long-running background jobs with $5 free compute/month. Both handle retries, scheduling, and durable execution. Hosting alternatives →

Durable / transactional workflows → DBOS

1M service calls/month with exactly-once execution guarantees. Ideal for payment processing, multi-step workflows, and anything where partial failure is unacceptable. Open-source DBOS Transact library available for TypeScript, Python, Go, Java, Kotlin.

Stateful edge computing → Cloudflare Durable Objects

Strongly consistent state at the edge with WebSocket hibernation. 100K requests/day on the free Workers plan. Unique capability — no other edge platform offers per-object consistency guarantees.

Hidden Costs and Gotchas

Serverless free tiers have several non-obvious costs that can surprise you. Here’s what to watch for.

API Gateway fees (AWS Lambda)

AWS Lambda invocations are free, but exposing them via API Gateway costs $3.50 per million requests (REST API) or $1.00 per million (HTTP API). For 1M Lambda invocations/month, the API Gateway bill can exceed $1-3.50. Google Cloud Functions and Azure Functions include HTTP triggers at no extra cost — a significant advantage over Lambda for HTTP workloads.

Egress charges

Most serverless platforms charge for outbound data transfer beyond free tiers. AWS: 100 GB free (first 12 months), then $0.09/GB. GCP: 1 GB/month free (North America). Azure: 100 GB/month free. Cloudflare Workers: Unlimited egress (no bandwidth charges). Deno Deploy: 20 GiB/month free. For bandwidth-heavy workloads, Workers’ unlimited egress is a major differentiator.

Cold start mitigation costs

AWS Lambda Provisioned Concurrency eliminates cold starts but costs $0.015/GB-hour — not free tier eligible. Azure Functions Premium keeps instances warm but starts at ~$0.173/vCPU-hour. Edge platforms (Workers, Deno Deploy) avoid this entirely with V8 isolate architecture — cold starts under 5ms without any warm-up configuration.

Vercel credit-based pricing complexity

Vercel restructured to a credit-based model in 2026. The Hobby plan includes 4 hours Active CPU and 360 GB-hours Provisioned Memory — but understanding how credits map to actual usage requires monitoring the Vercel dashboard. Serverless functions and edge functions consume credits differently, and overage charges apply automatically. Vercel details →

Workers CPU-time limit per invocation

Cloudflare Workers free tier limits each invocation to 10ms CPU time. This is CPU time, not wall-clock time, so I/O waits don’t count — but CPU-intensive tasks (heavy JSON parsing, image processing, crypto) can hit this limit. Paid Workers ($5/month) increases to 30 seconds CPU time per invocation. For most API workloads, 10ms CPU is sufficient.

Pricing Change Timeline

Recent serverless-related pricing changes from our tracker. See the full timeline for all 554 tracked changes.

Date Provider Change Impact
effective Oct 7, 2026 AWS AWS Proton end of support. Infrastructure-as-code provisioning service being retired. All environments, services, and templates will stop working. Source ↗ MEDIUM
effective Sep 30, 2026 AWS AWS App Mesh end of support. Service mesh for ECS/EKS being fully retired. All configurations, virtual nodes, and routes will stop functioning. Source ↗ MEDIUM
discovered Sep 7, 2026 · effective date unknown AWS Activate The AWS Activate program now offers up to $200,000 in credits, an increase from $100,000 for VC/accelerator-backed startups. The description of credits for AI startups building on custom silicon is no longer present. Source ↗ MEDIUM
discovered Sep 7, 2026 · effective date unknown Vercel The Hobby plan still exists, but several limits have been adjusted or are now pay-as-you-go. Edge Requests are now 1M included / month, then starting at $2 per 1M. Fast Data Transfer is 100 GB / month included, then starting at $0.15 per GB. Blob Storage is 1 GB / month included, then $0.023 per GB. Image Transformations are 5K / month included, then starting at $0.05 per 1K. Function Invocations are 1M / month included, then starting at $0.60 per 1M. Active CPU is 4 hours / month included, then starting at $0.128 per hour. Provisioned Memory is 360 GB-hrs / month included, then starting at $0.0106 per GB-hr. Source ↗ HIGH
discovered Sep 5, 2026 · effective date unknown Cloudflare Queues The free tier now offers 10,000 operations/day, and the paid tier offers 1,000,000 operations/month. Message retention on the free tier is 24 hours (non-configurable), while the paid tier offers 4 days default, configurable up to 14 days. The pricing model is now based on operations (64KB chunks of data written, read, or deleted) with additional costs per million operations beyond the included allowance. Source ↗ HIGH
discovered Sep 3, 2026 · effective date unknown LambdaTest The pricing for Test Manager Premium has changed to $49/month billed annually. KaneAI Starter is now $17/month billed annually. Source ↗ MEDIUM
discovered Sep 3, 2026 · effective date unknown Google Cloud Pub/Sub The Always Free tier of 10 GiB messages/month still exists, but it only applies to the Message Delivery Basic SKU. Throughput costs for BigQuery and Cloud Storage subscriptions, as well as import topics, are not free even within the first 10 GiB. Source ↗ MEDIUM
discovered Sep 3, 2026 · effective date unknown DBOS The pricing for DBOS Pro and DBOS Teams has been updated. DBOS Pro is $99/month for 2 user seats, 3 apps, and 1M checkpoints. DBOS Teams is $499/month for 10 user seats, 10 apps, and 10M checkpoints. Source ↗ MEDIUM
discovered Sep 2, 2026 · effective date unknown Cloudflare Durable Objects The free plan now only supports Durable Objects with SQLite storage backend. Also, the free tier limits for requests and duration have changed. Source ↗ HIGH
discovered Sep 1, 2026 · effective date unknown Cloudflare Workers AI The pricing has been restructured. While a free tier of 10,000 Neurons per day still exists, the pricing is now more granular and based on per-model unit pricing, billed in Neurons at $0.011 / 1,000 Neurons. Some models now require a paid plan or AI Gateway credits. Source ↗ HIGH
discovered Aug 28, 2026 · effective date unknown Deno Deploy Egress bandwidth for the free tier is now 20GiB, down from 100 GB. KV storage remains at 1 GiB, but the free tier now includes 1,000,000 KV read units and 500,000 KV write units. CPU time is now 10 hours, down from 15 hours. Revision storage is 10 GiB. Source ↗ HIGH
discovered Aug 28, 2026 · effective date unknown Inngest The event limit has increased from 100K to 500K, and the span data ingested limit has increased from 100MB to 500MB. The number of users is no longer listed as a limit for the Hobby tier. Source ↗ LOW

Data Source

Powered by AgentDeals. The tables on this page were compiled by hand from official pricing pages and have not been re-checked since. We track pricing changes across all major serverless providers. The pricing changes we track are updated continuously; the tables above are not.

Related guides: Cloud Free Tier Comparison · Hosting Alternatives · Free DevOps Stack · Free Tier Risk Index

Query serverless pricing programmatically via our MCP tools — compare serverless providers, track pricing changes, or plan your infrastructure stack from your AI coding assistant.

A source check reads each cited page for that service's name and a price; the figures in the tables above are from our own records.

Get this data in your AI editor

Compare serverless free tiers, track pricing changes, and plan your infrastructure stack — all from your AI coding assistant.

claude mcp add agentdeals -- npx -y agentdeals

Frequently Asked Questions

Which serverless free tier is most generous?

Free tier generosity varies by use case. Some providers offer more storage, others more compute or API calls. Our comparison table above shows exact limits side-by-side so you can evaluate based on what matters most for your workload.

How do serverless free tiers compare on limits?

Each provider structures free tier limits differently — some cap storage, others cap requests or compute hours. Our comparison table provides exact numbers for each provider. Check the growth cost analysis section to understand what you'll pay when you exceed free tier limits.

Related Comparisons

Related Guides

Explore all 1,547 developer tool deals → Browse the full index or connect via MCP