Skip to content

Stack AI vs Vellum (2026): which should you pick?

As of Nov 12, 2026, Vellum grades higher than Stack AI: 2.6 against 2.2 out of 4.0.

Checked Oct 9, 2026 and Nov 12, 2026 · graded Oct 10, 2026 and Nov 12, 2026 by ToolGPA

Vellum grades higher on the free plan and platforms. They get the same grade on value for money, features and ease of use.

Price, platforms and features side by side. The table says where each price was read and when.

Stack AI

Enterprise platform for building and deploying AI agents

Free plan

Vellum

Personal AI assistant with long-term memory that acts across apps

Free plan · from $30/mo

Which one is for you

Pick Stack AI if

  • Fits if you want to build one or two agents and try them free
  • Makes sense if you plan to ask for an enterprise deployment later

Look elsewhere if

  • Tight if you work in a team, as Free is 1 seat and 2 projects
  • Not for you if you want a self-serve paid plan, because none is published
  • Watch out: SSO, access control, VPC and on-prem are Enterprise only
  • Watch out: the homepage says Asana acquires StackAI, so ownership and plans may change

Pick Vellum if

  • Fits if you want one assistant that keeps your preferences across devices and chat apps
  • Works for people who want to read the code or host the open-source runtime themselves

Look elsewhere if

  • Skip it if you want free to mean free, as every reply, search and image draws on credits at $1 each
  • Not for you if you bank what you don't use: included usage doesn't roll over and bought credits expire after 12 months
  • Tight if you're on Windows or Linux, since the Windows app is a dev build and there's no Linux desktop client
  • Watch out: the iPhone and Android apps are clients for the cloud assistant and can't run it on the phone
  • Pricey if you want a custom subdomain, static IP or priority support, which start at Super for $100 a month

How we grade them

2.2GPAStack AIFair2.6GPAVellumGood
Value for money
COnly Free and a custom Enterprise plan are listed, so there is no published paid price to set against the category.
CMighty is $30 a month, right on the line where the dearest quarter of agent tools here begins, for a Small machine, 10 GiB of storage and a monthly usage allowance whose size the listing never states. Model usage is extra at $1 a credit. Super is $100 and Ultra $200.
Free plan and trial
CFree allows 500 runs a month, 2 projects and 1 seat, with community support on Discord. You will hit those caps.
BBase is free forever with no card, and the MIT runtime runs on your own Mac or Linux box for nothing. But free covers the hosting, not the thinking: every reply, search and image draws on credits at $1 each. Self-host with your own model key and one person could live on it.
Features
CA flow builder, knowledge bases and data loaders, and several ways to ship an agent. SSO, VPC, on-prem and BAA sit on Enterprise only.
CA knowledge base of facts and preferences it recalls later, skills across Google Workspace, Slack, Notion and GitHub, Allow or Deny prompts on risky actions, computer control and a form-filling Chrome extension. But the docs name two gaps: the Windows app is a dev build, and the phone apps can't run it locally.
Ease of use
CNo store rating on record, and building an agent means setting up flows and data.
CNo rating on record. The cloud default means nothing to host, but you still connect OAuth integrations, prepay credits and build trust rules before the permission prompts stop interrupting. The docs also call the Windows app a dev build.
Platforms and integrations
BWeb plus a browser extension, with Slack, Google Drive and Notion listed.
AWeb, Mac, Windows, iPhone, Android, a Chrome extension and a command line, plus Telegram, Slack, email, voice and phone calls, with integrations for Gmail, Calendar, Slack, Notion, Linear and GitHub. No Linux desktop client, and the phone apps are cloud clients.
Best forbuilding a first AI agent on a free planone assistant with memory on every device

The GPA is the average of the five grades: an A is 4 points, an F is none. The full reasons are on each tool's page: Stack AI and Vellum. How we grade

Where they differ

  • Vellum is open source and self-hostable, according to its own site.
  • Vellum also covers chat assistant.
In commonFree planWebBrowser extensionWorkflow automationAI agents

At a glance

Stack AIVellum
PricingFree planFree plan
Starts atNot published$30 per month (Mighty)
Free plan500 runs per month, 2 projects, 1 seat and community support on Discord.Base plan, free with no card: a Small cloud machine (1 vCPU, 2 GiB RAM) and 6 GiB of storage. Model usage, web search and image generation draw on pay-as-you-go credits. Self-hosting the runtime is also free.
Free trialNone listedNone listed
Web
macOS
Windows
iOS
Android
Browser extension
Made byStackAIVellum
Price read fromstackai.comvellum.ai
CheckedNov 12, 2026Oct 9, 2026

Worth knowing

Stack AI

  • Only a Free plan and a custom Enterprise plan are listed; there is no published self-serve paid tier.
  • The free plan has 1 seat, 2 projects and 500 runs a month.
  • SSO, access control, VPC, on-prem and BAA are Enterprise features.
  • Its homepage says Asana acquires StackAI, so plans and ownership may change.

Vellum

  • The Base plan is free to host, but model usage, web search and image generation are paid for in prepaid credits at $1 per credit.
  • Included monthly usage does not roll over, and pay-as-you-go credits expire twelve months after purchase.
  • The docs describe the Windows app as a dev build tied to Vellum's development environment, and no Linux desktop client is shipped.
  • The iOS and Android apps are clients for a cloud assistant; they do not run a local assistant on the phone.

Key features

Stack AI

Agent and flow builder
The comparison table lists Agents and Flow modules, with Python and JavaScript code steps for custom logic.
Multi-modal inputs
Agents take text, image, audio and video inputs and return text, image or audio, with vision and speech modules.
Knowledge bases and data loaders
Load documents (PDF, Word, PowerPoint), web scrapes, Google Drive and Notion content into a knowledge base.
Ways to ship an agent
Deploy through a REST API, a Chrome extension or a Slack bot; custom domains and analytics are listed.
Security and deployment options
Enterprise offers SSO, access control, VPC and on-premise deployment, and the page claims SOC 2, HIPAA and GDPR compliance.

Vellum

Long-term memory
The assistant keeps facts, preferences and people in a personal knowledge base and recalls them across conversations. On your own machine the workspace is plain-text files plus a SQLite database.
Skills for your work tools
Built-in skills and OAuth integrations cover Google Workspace, Outlook, Slack, Notion, Linear, GitHub, HubSpot, Asana and Todoist. More skills come from a catalog or are written yourself.
Permission prompts
Sensitive actions such as file access, shell commands or emailing a new contact show an Allow or Deny prompt with a risk badge until you create a trust rule for them.
One assistant on every channel
The same identity, memory and schedules are available from the web, Mac, Windows, iOS and Android apps, a CLI, Telegram, Slack, email, voice and phone calls.
Computer and browser control
The desktop app adds voice input and computer control, and a Chrome extension lets the assistant read open tabs, fill in forms and act on the page in front of you.
Cloud, local or self-hosted
Run it in Vellum Cloud, locally on a Mac or Windows PC, or self-hosted on a Linux host or your own Google Cloud project, bringing your own model API key.

Plans

Stack AI

  • Free plan$0

    500 runs per month, 2 projects, 1 seat and community support on Discord.

  • EnterpriseCustom

    Custom runs and seats, unlimited projects, dedicated infrastructure, SSO, VPC or on-prem deployment.

Vellum

  • Base$0

    Free forever: a Small cloud machine (1 vCPU, 2 GiB RAM), 6 GiB of storage and managed model access paid for with pay-as-you-go credits.

  • Self-hosted$0

    The open-source runtime on your own Mac, Windows PC, Linux host or GCP project, with your own model API key and no platform or machine fees.

  • Mighty$30

    per month

    Small machine, 10 GiB of storage and a monthly Mighty Usage allowance. Leaves out the custom subdomain, static IP and priority support.

  • CustomFrom $50

    per month

    Pick a machine size, storage tier and optional usage bundle yourself. The minimum is a Medium machine with 10 GiB and no usage bundle.

  • Super$100

    per month

    Medium machine (2.5 vCPU, 5 GiB), 30 GiB of storage and Super Usage, plus an assistant email, custom subdomain, static IP and priority support.

  • Ultra$200

    per month

    Large machine (4 vCPU, 8 GiB), 60 GiB of storage and Ultra Usage, with the same assistant email, subdomain and support as Super.

Common questions

Which is cheaper, Stack AI or Vellum?

As of Nov 12, 2026, Stack AI publishes no starting price, and Vellum starts at $30 per month (Mighty).

Which has a free plan, Stack AI or Vellum?

As of Nov 12, 2026, both Stack AI and Vellum have a free plan.

Which platforms do Stack AI and Vellum run on?

As of Nov 12, 2026, only Vellum lists macOS, Windows, iOS and Android apps. Both list a web app and a browser extension. A platform missing from a listing is one its maker's pages did not mention.

More comparisons

How this page was made

  • The verdict at the top follows a fixed rule. When both tools are in the same category and either is in our top three for it, the verdict follows that ranking: the higher-placed tool is our pick. The top three weigh who each tool suits as well as its grades, so a pick can go against the GPA. Otherwise the page does not pick one and says which grades higher: the higher GPA first; on a tie, the one with a free plan, then the one that starts cheaper in dollars a month, counted the same way for both; if they are still level, what the choice depends on.
  • The grades are our own, from each tool’s review or report card. The GPA is their average.
  • Prices, plans and platforms come from each tool’s listing. The table above says where each price was read and when.
  • The differences and the answers to the questions are worked out from those listings.
How we review and grade