Netlify, the web hosting platform founded in 2014, has just landed on AIxploria as a new AI tool. The directory added its card at 03:22 UTC on 28 September 2026, filing it under Developer Tools and No Code/Low Code. The card’s blurb says you can “create and deploy applications from a command prompt, Git, or the terminal,” with coding agents from Claude, Codex and Gemini working “directly within your projects.”
Every claim on the card is true. The problem is what it leaves out. The card labels it “Freemium”, but the company’s own documentation warns that on the free plan “a single prompt can use most of your available credits.” It also says that once a balance is used up, every site on the account is paused and visitors see a “Site not available” page.
We checked the card against Netlify’s documentation, pricing, changelog and status page. Here is what its AI features do, what they cost, what the card’s links really are, and what to set up before your first agent run. For a wider view of the field, see the AI models and tools hub.
Table of contents
- The Netlify Card at a Glance
- What the Platform Is, and Why It Is “New”
- Netlify Card Claims Checked
- How Agent Runners Work
- The Netlify Free Plan Warning the Card Leaves Out
- What Netlify AI Costs Compared With Going Direct
- The Links on the Card
- The Alternatives Strip
- The Rank Badge on the Card
- A Busy Week Before the Listing
- Should You Try It for AI Building?
- Frequently Asked Questions About Netlify
- References
The Netlify Card at a Glance
AIxploria gives each listing a blurb, tags, a rating and a rank badge. Most model cards also carry a long written review. This one does not.
The basics
The card sits at aixploria.com/en/netlify-ai/. Its tags are “AIxploria Selection”, “Developer Tools”, “Latest AI” and “No Code/Low Code”. It is marked as a Verified Tool and priced as Freemium. The page title promises “26 Alternatives”, but the page shows eight.
Eight votes
The card shows a 4.5 rating. The page’s structured data shows that figure comes from eight votes. The card also shows 67 upvotes. The MiMo-V2.6 card, added 26 minutes later, had seven votes.
No review
The card has no written review, no pros and cons, and no FAQ. Its text is a 34-word blurb and a one-sentence page description. That leaves the card’s price label, badge and links carrying most of the weight.
The rank badge
The card is badged “#14 in Developer Tools”. We checked that number against the category page, and the result is covered below.
| Card field | Value on 28 September |
|---|---|
| Published | 03:22 UTC, edited at 03:27 |
| Rank badge | #14 in Developer Tools |
| Rating | 4.5 from 8 votes |
| Upvotes | 67 |
| Price label | Freemium |
| Written review | None |
| Useful links | Official Website, Signup, Drop |
| Alternatives shown | 8, of “26” in the page title |
What the Platform Is, and Why It Is "New"
The company is not new. It was founded in 2014 and is based in San Francisco. It describes itself as “home to millions of developers and thousands of enterprises.” What is new is how much of the platform is now built around AI agents.
From hosting to agents
The company made its name hosting websites that are built once and served worldwide, with serverless functions for the dynamic parts. Its pitch has long been that small teams can ship without specialist DevOps expertise. Its own summary for AI tools now calls it the platform for “modern web projects — and the agents that build them.”
The Bolt.new connection
The platform was already part of the AI building boom before this listing. In March 2025 it announced that more than one million sites built with the AI app builder Bolt.new had been deployed on it in five months. AIxploria has listed Bolt.new since November 2024. The platform those sites ran on arrived on the directory almost two years later.
Agent Experience
The company, led by co-founder and chief executive Mathias Biilmann, coined the term “Agent Experience” in 2025. The idea is that AI agents are becoming users of software in their own right, so platforms need to work well for them. It publishes a guide for agents, an MCP server that lets coding agents deploy sites, and ready-made “skills” files that teach agents how the platform works.
Netlify Card Claims Checked
We compared each claim in the card’s blurb, page description and tags with the company’s documentation.
| Card says | The documentation says | Verdict |
|---|---|---|
| Deploy from a prompt | “Deploy from AI, Git, or API”; Agent Runners start from a prompt | Correct |
| Deploy from Git | Yes, but Agent Runners only work with GitHub | Incomplete |
| Deploy from the terminal | Its command-line tool | Correct |
| Code agents | Agent Runners, on credit-based plans only | Correct |
| Serverless functions | Functions, edge functions, background functions | Correct |
| Databases | Its own managed Postgres database | Correct |
| Storage | Blobs storage, still labelled Beta | Incomplete |
| Authentication systems | Its Identity service, on all credit-based plans | Correct |
| Pre-release previews | Unlimited Deploy Previews on every plan | Correct |
| Claude, Codex and Gemini | Plus OpenCode, with open models such as Kimi, DeepSeek and GLM | Incomplete |
| Freemium | Free plan: 300 credits a month, hard limit | Incomplete |
| No Code/Low Code | “No technical background needed” for Agent Runners | Correct |
The score
Of 12 claims, eight are correct and four are correct but incomplete. None is wrong. The incomplete ones matter most to a first-time user, because they cover which Git host works, what “free” means and which agents are on offer.
How Agent Runners Work
Agent Runners is the feature behind the card’s “code agents” line. It lets you type a request in the dashboard and have an AI coding agent change your project.
Four agents
Agent Runners supports Claude Code, OpenAI Codex, Google Gemini and OpenCode. OpenCode offers open models, served through OpenRouter. The company says it only routes those requests to providers with a zero data retention policy, meaning they do not store your prompts or outputs.
Build and Ask
Each run uses one of two modes. Build mode changes your code and creates a Deploy Preview for you to check. Ask mode answers questions about your project without changing anything. Your live site does not change until you choose to publish.
What the agent can see
The pitch is context. An agent running on the platform can see your code, environment variables, build settings and deployment pipeline. The company says that lets it “wire up a serverless function, configure auth, or set up a form without extra setup.”
Who can use it
Agent Runners works only on the credit-based plans: Free, Personal and Pro. Enterprise customers must ask their account manager. Team owners, developers, internal builders and publishers can start runs.
The limits the card skips
If your project is connected to a Git repository, Agent Runners only works with GitHub. Projects on GitLab, Bitbucket or Azure DevOps cannot use it. It also does not work with the Split Testing feature, and each team can only run a limited number of agents at once.
The Netlify Free Plan Warning the Card Leaves Out
This is the most important section for anyone who clicks the card’s Signup link. The Free plan is real, and it costs nothing. But its limits work differently from most free tiers.
300 credits, hard limit
The Free plan gives 300 credits a month, with a hard limit and no way to buy more. Everything uses credits: a production deploy costs 15, bandwidth costs 20 per GB, compute costs 10 per GB-hour and web requests cost 2 per 10,000.
What AI costs in credits
The platform converts AI model spending at 180 credits per US dollar. So the Free plan’s entire monthly allowance equals $1.67 of AI model usage, if nothing else uses credits that month.
One prompt can use most of it
Netlify’s own Agent Runners guide says it plainly: “especially on a free plan, a single prompt can use most of your available credits.” It ranks each model’s cost from one to five and suggests cheaper models for iteration.
An empty balance pauses every site
Netlify’s billing guide says: “Once your credit balance is completely used up, all of your web projects (sites/apps) are paused and visitors to your web projects will find a Site not available page.” On the Free plan, the balance comes back only when the monthly credits reset or you upgrade.
AI is on by default
The documentation says AI features are “on by default for all Credit-based Free, Personal, and Pro plans.” A team owner can turn them off, and teams can set an AI credit usage limit. That limit covers Agent Runners but not the AI Gateway, its service for calling AI models from your own code.
| What 300 credits buy on their own | Amount |
|---|---|
| Production deploys (15 credits each) | 20 deploys |
| Bandwidth (20 credits per GB) | 15 GB |
| Compute (10 credits per GB-hour) | 30 GB-hours |
| Web requests (2 credits per 10,000) | 1.5 million requests |
| AI model usage (180 credits per $1) | $1.67 |
A worked agent run
Suppose one agent run reads one million input tokens and writes 50,000 output tokens. That is a modest job for a coding agent that re-reads a project on each step. On Claude Opus 5.5, at $4 and $20 per million tokens, the run costs $5, or 900 credits. That is three months of the Free plan’s allowance, before compute.
The same run on a cheaper model
On Gemini 3.8 Flash, at $0.75 and $3.75 per million tokens, the same run costs about $0.94, or 169 credits. That is still more than half of the Free plan’s month, which is why the documentation steers new users toward cheaper models.
What Netlify AI Costs Compared With Going Direct
The platform passes AI model prices through at the providers’ list rates, then converts the dollar total into credits. The price of those credits depends on your plan.
How we worked it out
Each dollar of model usage costs 180 credits. On the Pro plan, credits cost $10 per 1,500, so 180 credits cost $1.20. On a Personal credit pack, credits cost $5 per 500, so 180 credits cost $1.80. The Personal plan’s included credits sit between the two.
What the premium buys
The company says the rate is “designed to make using AI features like Agent Runners and AI Gateway competitive with using AI providers without Netlify.” The extra cost buys no API keys to manage, one bill, and the agent’s access to your project’s settings. Whether that is worth 20% to 80% depends on how much you use.
Plans at a glance
| Plan | Price | Monthly credits | When credits run out |
|---|---|---|---|
| Free | $0 | 300 | Hard limit; sites paused |
| Personal | $9 a month | 1,000 | Optional auto recharge, 500 for $5 |
| Pro | From $20 a month | 3,000 | Optional auto recharge, 1,500 for $10 |
| Enterprise | Custom | Unlimited | Not applicable |
Database limits
Its managed database is included on every credit-based plan. The Free plan allows three databases and 20 active branches, with seven days of backups. Pro allows 50 databases and 300 branches, with 30 days of backups.
The Links on the Card
The card has three useful links. Two of them are not plain links.
Signup and Drop
The “Signup” link goes to join.netlify.com, which redirects to the sign-up page with tracking tags that include utm_source=partnerstack and a ps_partner_key. The “Drop” link does the same for Drop, the drag-and-drop publishing page. Both set a tracking cookie that lasts 90 days.
The partner programme
PartnerStack is the service the company uses to run its partner programme. Its partners page offers “up to 20% revshare for agencies, creators, and developers” who refer customers. The two links carry a partner key, so sign-ups through them can be attributed to a partner account.
What this does and does not show
A partner link does not make the card wrong, and it does not tell us how much, if anything, AIxploria earns. But the card does not say the links are tracked. By contrast, the MiMo-V2.6 card published the same morning links straight to Hugging Face and Xiaomi’s own pages.
What Drop does
Drop is one of the platform’s oldest features. You drag a project folder onto the page and get a live address in seconds. The docs say deploys under 50 MB work best. A project dropped without signing in is protected with a temporary password until you claim it.
The Alternatives Strip
The card lists eight “alternatives”. They say more about AIxploria’s categories than about the platform’s real rivals.
| Listed alternative | What it is | Hosts what you build? |
|---|---|---|
| Replit AI | Browser coding platform with an agent | Yes |
| Hugging Face | AI model and dataset hub | Only demo apps |
| MiniMax M2.5 | An AI language model | No |
| Playground OpenAI | OpenAI’s model testing console | No |
| OpenClaw | Self-hosted personal assistant | No |
| Browser Use | Browser automation for agents | No |
| Windsurf Editor | AI code editor | No |
| Claude Code | Anthropic’s coding agent | No |
One alternative runs inside the platform
Claude Code appears in the strip as an alternative, but it is one of the four agents that run inside Agent Runners. Choosing this platform does not mean giving up Claude Code. If you use it, read our report on how Claude Code deleted 48,000 files in 103 seconds, which shows why a review-before-publish step matters.
Missing rivals AIxploria does list
AIxploria lists several tools that do what these AI features do. Lovable is badged “#2 in No Code/Low Code”, Bolt.new “#9” and v0 “#27”. None of them is in this card’s strip. AIxploria’s own Bolt.new card even lists “Built-in GitHub export and Netlify deployment” among its pros.
The Rank Badge on the Card
The card says “#14 in Developer Tools”. We walked the first three of the category’s 22 pages, 36 cards in all.
Not on the first three pages
The card does not appear in the first 36 positions. Slot 14 holds Replit AI. Page four and beyond returned a Cloudflare challenge, so we cannot say where it does sit.
The badges mostly match
We opened 21 of the 36 cards and read their badges. Ten of those have Developer Tools as their main category. Nine showed their exact position. The tenth, Replit AI, sits in slot 14 but carries a “#8” badge.
| Slot on the category page | Card | Its own badge |
|---|---|---|
| 2 | Claude Code | #2 |
| 4 | Grok Build | #4 |
| 7 | Google AI Studio | #7 |
| 13 | KiloClaw | #13 |
| 14 | Replit AI | #8 |
| 20 | Google Antigravity | #20 |
| 27 | Gemini CLI | #27 |
Two cards point at slot 14
The new card’s badge claims slot 14, and the card sitting there, Replit AI, claims slot 8. Slot 8 holds Browser Use, whose own badge is for a different category. Neither card is where its badge says.
Where the card has spread
The Latest AI category’s “New AI tools” strip lists the card second, behind MiMo-V2.6. It has no mentions on the Top 100 page or the Ultimate List. Cards added on 24 September still have none there after four days.
A Busy Week Before the Listing
The card arrived at the end of a busy week for the company. Four events from its changelog and status page are worth knowing.
Two major outages
The status page records a “major” incident on 21 September, “Errors accessing Netlify-hosted sites”, that ran from 15:12 to 16:37 UTC. A second major incident with the same description ran from 19:01 to 19:25 UTC on 24 September. A smaller CDN problem, caused by DNS resolution, fell between them.
A Next.js security warning
On 22 September the company warned customers about a critical cybersecurity flaw in Next.js’s image-generation tool, CVE-2026-94545. It said the impact on its platform is limited to a crashed function call, but that “active exploitation could increase your function costs.” Its advice was to upgrade to Next.js 15.5.26 or 16.3.6 and redeploy.
New models for agents
The same day, it added Claude Opus 5.5, GPT-6 Sol and GPT-6 Luna to Agent Runners and the AI Gateway. We checked the GPT-6 cards in New AI Tool: GPT-6 Sol and Luna Has Just Landed on AIxploria.
An activity feed
On 24 September the company added an activity feed to each project’s overview page. It shows production versions, previews and agent runs in one list, including what each teammate asked the agents to do.
Should You Try It for AI Building?
For small web projects, Netlify’s AI features are a sensible way to go from prompt to live site. Every change gets a preview, and nothing goes live until you say so. The question is whether the credit model suits you. Our AI coding assistants guide covers the agents it runs.
Good fits
- Marketing sites, landing pages and prototypes, where agents make small, well-defined changes.
- Non-technical team members who need to make content updates with a preview before anything ships.
- Teams already on GitHub who want an agent that can see their real build settings.
Poor fits
- Projects on GitLab, Bitbucket or Azure DevOps, which Agent Runners does not support.
- Heavy agent use on the Free plan, where one large prompt can take a month’s credits and pause your sites.
- Anyone who wants the cheapest possible AI tokens, since its credits cost 20% to 80% more than going to the provider directly.
A safe first week
If you sign up, set an AI credit usage limit first, before any agent runs. Try Ask mode before Build mode, since it changes nothing. Pick a cheaper model for small edits. If the account hosts a site you care about, put it on a paid plan, where running out of credits does not have to mean going offline. For larger builds, our MVP development services can help.
Frequently Asked Questions About Netlify
Is Netlify free?
Netlify has a free plan with 300 credits a month and a hard limit. You cannot buy more credits on it. When the credits run out, all your sites are paused until the next month or until you upgrade.
Which AI agents are supported?
Agent Runners supports Claude Code, OpenAI Codex, Google Gemini and OpenCode. OpenCode gives access to open models such as Kimi, DeepSeek and GLM.
How much does the AI cost?
The platform charges the model provider’s list price, converted at 180 credits per dollar. Depending on your plan, that works out at $1.20 to $1.80 for each dollar of model usage.
Are the Signup links on AIxploria affiliate links?
The Signup and Drop links on the card carry PartnerStack tracking tags and a partner key. The company runs its partner programme, which offers revenue share on referrals, through PartnerStack. The card does not say the links are tracked.
Is the AIxploria rank badge accurate?
No. The card says #14 in Developer Tools, but the card was not in the category’s first 36 positions, and slot 14 holds Replit AI.
References
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.
