Vellum
Open-source personal AI assistant with memory, from the former LLMOps platform
Overview
Vellum is an open-source (MIT) personal AI assistant that remembers you across conversations and acts for you across email, calendar, files, the browser and tools such as Slack and Notion. It runs locally or in Vellum Cloud, reachable from Mac, Windows, iOS, Android, web and messaging apps. Until early 2026 the New York company sold an LLM app-development platform.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 12 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- Your first tasks are on us
- Small machine (1 vCPU, 2 GiB) and 6 GiB storage
- Pay-as-you-go credits; managed LLM credentials
- Small computer, 1 vCPU / 2 GiB
- 10 GB storage
- Mighty usage allowance included
- Medium computer, 2.5 vCPU / 5 GiB
- 30 GB storage
- Super usage allowance; assistant email and subdomain
- Large computer, 4 vCPU / 8 GiB
- 60 GB storage
- Ultra usage allowance; assistant email and subdomain
- Choose machine, storage and an optional usage bundle
- Plus the $10 platform fee
- No platform fee or computer charges
- You control data, keys and uptime
Use cases
What you can produce with Vellum
- An always-on assistant for email, calendar, files, browsing and recurring tasks
- Long-term memory kept as a browsable, correctable wiki of concepts
- Permission prompts for high-stakes actions until you mark them trusted
- Access from Mac, Windows, iOS, Android, web, Telegram, Slack or a terminal
- Models switchable across OpenAI, Anthropic, Gemini or Ollama open weights
- MIT-licensed code to self-host free, or Vellum Cloud
- Pay-as-you-go credits at $1 each, passed through at model cost
Vellum is now a personal assistant, not the eval-and-workflow platform its name still evokes for developers. The pitch is ownership: the code is MIT-licensed, the assistant runs locally by default, and it keeps a browsable long-term memory you can correct or switch off. It drafts email, manages calendars, browses and runs recurring jobs, asking permission for risky actions until you trust them.
Self-hosting is free; cloud plans run from $30 to $200 a month plus pass-through model credits at $1 each. It is young, with releases every few days, so expect rough edges, and teams on the old LLMOps product should ask the vendor directly about its future.
What we found
Vellum is no longer the product most search results describe, and earlier versions of this page described the old one. Until early 2026 the New York company sold an LLM application development platform: prompt management, evaluations, visual workflows and deployments. Its last platform changelog entry is dated 4 February 2026, and the marketing pages for those products now redirect to the homepage, though the developer docs and SDKs remain online.
The website now sells an open-source personal AI assistant. The code sits on GitHub under the MIT licence, and the company's launch post, last updated 7 May 2026, says it has raised $25 million from backers including Dharmesh Shah, Arash Ferdowsi, Rebel Fund and Y Combinator. You give the assistant a name and a personality, and it works across email, calendar, files, the browser, Slack, Notion and Google Workspace, asking permission for risky actions until you mark them as trusted. It runs locally by default or in an isolated cloud container, and you can reach it from Mac and Windows apps, iOS, Android, the web, Telegram, Slack or a terminal. Models are switchable, including OpenAI, Anthropic, Gemini and open weights through Ollama.
Memory is the headline feature. The launch post describes a small set of always-loaded files plus a knowledge graph retrieved selectively; the current docs present it as a linked wiki of concepts you can browse, refine or switch off.
Pricing has two parts. Self-hosting is free. In Vellum Cloud, the free Base plan includes a small machine and 6 GiB of storage, and paid work is billed in credits at $1 each, which the company says it passes through at model-provider cost. Packages bundle compute, storage and a monthly usage allowance: Mighty at $30, Super at $100 and Ultra at $200 a month, with a Custom build from $50.
Releases are frequent; version 0.12.6 shipped on 29 September 2026. Customers of the old platform will find nothing on the current site about its future.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
About this listing
This entry was compiled from publicly available data including Vellum's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Vellum unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Vellum directly →
Spotted something out of date? Suggest an update →
Alternatives to Vellum
More in Productivity
