Teams look beyond AWS Lambda when the bill is hard to forecast, when they want code that runs outside one cloud, or when a smaller platform covers what they actually deploy. Lambda sells several billing models side by side, and its strongest feature, wiring into other AWS services, matters less once the rest of the stack lives elsewhere.
Short answer: Looking for a replacement for AWS Lambda? Nuclio, Bunny Edge Scripting and Wasmer Edge come first in our serverless platform ranking. Challengers take places 1 to 3 here. Big brands follow in the order of our category ranking. Start with Nuclio if functions must run on servers you control, outside any cloud. The 11 entries that follow are all ranked in Best Serverless Platforms in 2026; each one explains what sets it apart from AWS Lambda.
Vendors can pay for more visibility on this website, for example a higher place in a ranking or more room on a page. Commercial relationships can also affect which products we feature and how prominently. Every vendor link is marked the same, whatever the arrangement behind it. How that works.
Pricing: Published per-request and per-GB-second rates with a monthly free tier; other models for web functions and managed instances
Established in: Not stated by the vendor
Built for: Teams already building on AWS that need functions triggered by its services
Not for: Buyers who want one flat price per function
01
Why buyers look past AWS Lambda
01
Many billing models to estimate
The Lambda pricing page lists Event Functions at $0.20 per million requests plus $0.0000166667 per GB-second, then separate rate tables for Provisioned Concurrency, SnapStart, Managed Instances, Web Functions, MicroVMs, durable functions, response streaming and Lambda@Edge. A monthly figure needs a decision about which model applies, and the free tier of 1,000,000 requests and 400,000 GB-seconds covers only part of them.
02
Limits that shape the design
The documentation sets a 15-minute timeout for functions, 6 MB for synchronous request and response payloads, a default of 1,000 concurrent executions per Region and new-account quotas that start lower until AWS raises them. Longer jobs, bigger payloads or sudden traffic from a new account therefore need a quota request or a different compute model.
03
The value sits inside AWS
The product page counts more than 220 native AWS integrations, with S3, DynamoDB, Kinesis, Kafka and SQS named as event sources. That helps when the data already lives in AWS. A team that keeps its database, queue or web front end elsewhere carries the account, permissions and billing structure of AWS for a function that calls out to other services.
your organisation standardises on Microsoft cloud rather than AWS
#14
03
What each one does differently
1
Nuclio
Not stated by the vendor · See profile
Nuclio is the only entry here that you install on your own machines. The code sits under an Apache-2.0 licence in its official GitHub repository and runs on Docker, Kubernetes, on-premises or across clouds. Its homepage lists no prices or limits and names no country, so support terms and the managed option need a direct conversation.
Switch from AWS Lambda if functions must run on servers you control, outside any cloud.
Scripts are written in TypeScript on the Deno runtime, with npm modules, and go live across 119 edge locations in under 100 ms. Billing counts requests and CPU time with a monthly minimum, and a 14-day trial is listed. Bandwidth is priced separately and the product page gives no figures for it. No country appears on the pages we read.
Switch from AWS Lambda if short scripts near the visitor matter more than deep cloud integrations.
The Hobby plan states its ceiling in numbers: 100 apps, 1,000,000 requests and 150 GB of bandwidth a month. Pro is billed per seat and adds backups, SSH access and cron jobs, with overage rates listed. The vendor says Wasmer can also run on infrastructure you own. Country and licence for Edge are not stated.
Switch from AWS Lambda if you want a free tier with published caps for small hosted apps.
Supabase Pte. Ltd. of Singapore bundles functions with a hosted database. Code runs on Deno with npm specifiers, database webhooks fire on INSERT, UPDATE and DELETE, and functions can sit close to the data. The free plan includes 500,000 invocations but pauses projects after a week of inactivity, and Pro includes 2 million.
Switch from AWS Lambda if your functions mostly react to rows in a Postgres-style database.
Containers rather than functions: Koyeb runs them in VMs on bare-metal servers in more than 50 locations, scales to zero and bills compute per second. The Pro plan adds a monthly fee and includes $10 of compute. A banner on the about page says Koyeb is joining Mistral AI, with no terms given, so ownership may change.
Switch from AWS Lambda if your workload is a container image and per-second billing suits it.
Deno Land Inc. of San Diego runs this host for Deno projects, with GitHub deploys and a preview domain for each pull request. Cron jobs are defined in code and OpenTelemetry tracing is added automatically. The free plan stops at 1M requests and 10 apps with community support only. A banner says Deno is joining Cloudflare, with no terms.
Switch from AWS Lambda if preview deployments per pull request are central to how you ship.
Modal aims at compute-heavy work instead of web glue: a function becomes an HTTPS endpoint that scales to zero, with per-second billing for CPUs and GPUs. The Starter base fee is $0 with monthly credit, while Team costs $250 a month plus compute. The terms name Modal Labs, Inc., a Delaware corporation, and no country.
Switch from AWS Lambda if you run GPU or batch jobs that outgrow Lambda's 15-minute window.
Netlify, Inc. of San Francisco ships functions with the site on every plan. Background functions run up to 15 minutes, scheduled functions take a cron expression and memory reaches 4 GB on credit plans. Usage is metered in credits at 10 per GB-hour, and the pricing page gives no invocation counts, so forecasts require converting credits first.
Switch from AWS Lambda if functions are a small part of a site you already host there.
Limits are printed per plan: duration up to 300 s on Hobby and 800 s on Pro, memory to 2 GB or 4 GB, 4.5 MB payloads and concurrency up to 30,000. Hobby usage is hard-capped and extra cannot be bought, so growth forces Pro, where invocations, Active CPU and memory appear as separate lines. The terms name Vercel Inc. without a country.
Switch from AWS Lambda if the front end already deploys there and you want stated limits.
Cloudflare, Inc. of San Francisco runs Workers in isolates rather than containers. The Free plan allows 100,000 requests a day with 10 ms of CPU per invocation; the paid plan has a minimum charge plus request and CPU-time rates. Each isolate has 128 MB, so memory-hungry code does not fit. Cron triggers are included.
Switch from AWS Lambda if light request handling near users suits you and 128 MB is enough.
Microsoft's counterpart offers Consumption, Flex Consumption, Premium and Dedicated plans. Consumption grants 400,000 GB-seconds and 1 million executions a month, and a $200 credit lasts 30 days. Native languages are C#, Java, JavaScript, PowerShell and Python. Linux Consumption retires in September 2028, and the contracting company's country is not stated.
Switch from AWS Lambda if your organisation standardises on Microsoft cloud rather than AWS.
Nuclio: It is published under Apache-2.0 and runs on Docker, Kubernetes or on-premises, though its homepage shows no prices or limits.
02
You react to database changes more than HTTP calls
Supabase Edge Functions: Database webhooks fire on INSERT, UPDATE and DELETE, and the free plan includes 500,000 invocations before projects pause after a week of inactivity.
03
You run GPU or batch compute that outgrows 15 minutes
Modal: Compute is billed per second and Starter has a $0 base fee with monthly credit, while Team costs $250 a month plus usage.
04
You want a free tier with stated usage caps for small apps
Wasmer Edge: Hobby allows 100 apps, 1,000,000 requests and 150 GB of bandwidth a month.
05
You deploy a front end and want limits printed per plan
Vercel Functions: Duration runs to 300 s on Hobby and 800 s on Pro, with 4.5 MB payloads and concurrency up to 30,000.
04
When AWS Lambda is still the right choice
Lambda remains the sensible choice when your data, queues and identity already sit in AWS. The 220-plus native integrations mean S3 uploads, DynamoDB streams or SQS messages can trigger code without glue. The monthly free tier is generous for small workloads, the 10,240 MB memory ceiling and 15-minute timeout cover most jobs, and a published request rate makes light usage cheap. Moving away costs you those triggers.
05
What moving off AWS Lambda involves
01
Event wiring does not travel
Triggers from S3, SQS, DynamoDB Streams or EventBridge are configured on the AWS side, so list each function with its event source before choosing a target. Anywhere else you must rebuild them as HTTP calls, webhooks, cron entries or queue consumers and decide who retries on failure.
02
Runtimes, packaging and permissions
Lambda code often assumes AWS SDK calls and IAM roles. Edge hosts based on Deno or isolates accept TypeScript and npm modules but not every native library, and some cap memory at 128 MB. Container-based hosts take your Lambda container image closer to as it is. Test cold starts and timeouts with real payloads.
03
Run both for a while
Keep the old function live and route a share of traffic to the new host. Compare the first full month of bills, because Lambda's request, duration, provisioned and data-transfer lines are priced differently from per-second or credit-based billing elsewhere.
06
Questions about replacing AWS Lambda
6 answers
What does AWS Lambda cost for ordinary functions?
The pricing page lists $0.20 per million requests and $0.0000166667 per GB-second for Event Functions on x86, with 1,000,000 requests and 400,000 GB-seconds free each month. Other models, such as Web Functions at $0.30 per million requests, have their own tables.
How long can a Lambda function run?
The documentation sets 900 seconds, or 15 minutes. Functions on Lambda Managed Instances invoked asynchronously or through most event source mappings can run up to 5,400 seconds (90 minutes). Anything longer needs another service or a different platform.
Which alternatives can I run on my own servers?
Nuclio, published under Apache-2.0, runs on Docker, Kubernetes or on-premises. Wasmer's page says Wasmer can run on infrastructure you own. The others in this list are managed platforms, and Koyeb and Modal run container-style compute.
Is there a free plan on the alternatives?
Wasmer Edge, Deno Deploy, Supabase, Netlify, Vercel and Cloudflare each list a free plan with caps of their own, such as 100,000 requests a day on Cloudflare or 500,000 invocations on Supabase. Azure Functions grants 1 million executions a month on Consumption.
Which options suit long or GPU-heavy jobs?
Modal bills compute per second and advertises GPU and CPU autoscaling. Koyeb runs containers and lists GPU concurrency. Netlify background functions reach 15 minutes and Vercel Pro allows 800 seconds.
Do the alternatives name a country?
Only some do. The records show Supabase in Singapore, Deno in San Diego, Netlify in San Francisco and Cloudflare in San Francisco. For several others, the pages we read name no country, so check the contracting entity before signing.
—
Sources
Checked 15 November 2026
AWS Lambda pricing: $0.20 per million requests, $0.0000166667 per GB-second, 1M requests and 400,000 GB-seconds free, Web Functions $0.30 per million, other billing models.
Founder of The Knowledge Engineers and of 2 Steps Ahead, an AI search visibility agency. Editorially responsible for every page on this site; research and drafting are AI-assisted.
Sebastiaan Smits wrote and checked this page, most recently on 15 November 2026. See how we review.