← Back to all sparks
K

Khoj

AI-ASSISTANTS
Velocity0.0

Open-source personal AI assistant that searches and reasons over your documents.

Khoj gave its assistant long-term memory, then started winding down its own cloud.

personal-ailong-term-memoryself-hostedcloud-deprecationobsidianmodel-support
Current state
The 2.0 beta line is doing two things at once. In February it gave Khoj long-term memories, moving the assistant from per-conversation context to knowledge that persists across sessions. Six weeks later it began deprecating Khoj cloud — a banner warning hosted users, and the subscribe option removed from the web app's settings page. The same release upgraded the web app to Next.js 15 and fixed the research agent stopping earlier than it should.
Where it's heading
Khoj is consolidating around the self-hosted product and the model layer beneath it, and stepping back from running the hosted service. Model support has been broadening steadily — Gemini Pro 3, Minimax M2, Nano Banana image editing, configurable fallback models with admin-set priority — while Stability AI image generation was dropped. The client work follows the same logic: Obsidian batch sync, file query filter autocomplete, reference copying, all aimed at people running their own instance against their own files.
Prediction
With subscriptions already disabled and a deprecation banner live, the next step for the hosted service is a shutdown date. The entries stop in late March 2026, so whether 2.0 has left beta since is not visible here.

Recent moves

  1. 4mo ago

    2.0.0-beta.28: two small web app fixes

    Repairs the web app for users who disabled memories without ever setting a chat model, and stops the deprecation banner appearing where it should not. Cleanup on the two changes that defined the preceding releases.

    View source ↗
  2. 4mo ago

    2.0.0-beta.27: page templates fixed for Starlette 1.0.0

    A single compatibility fix to the web app's page templates for Starlette 1.0.0. Dependency housekeeping, shipped hours after the previous beta.

    View source ↗
  3. 4mo ago

    2.0.0-beta.26 starts deprecating Khoj cloud

    ⚡ SPARK

    Hosted users get a banner about the coming cloud deprecation and can no longer subscribe from the web app's settings page. Alongside it, a Next.js 15 upgrade and fixes to the research agent stopping early and to Khoj ignoring pages it had already fetched.

    View source ↗
  4. 5mo ago

    2.0.0-beta.25 gives Khoj long-term memory

    ⚡ SPARK

    Long-term memories arrive, taking the assistant beyond a single conversation's context. The rest of the release is client work — Obsidian batch sync with progress and storage reporting, query filter autocomplete, and reference copying from the web app.

    View source ↗
  5. 6mo ago

    2.0.0-beta.20 adds Nano Banana image editing and model fallbacks

    Precise image editing arrives through Gemini Nano Banana, and administrators can configure multiple chat settings with priority so the server falls back between models. Defaulting tool calls to a fast model trims response time.

    View source ↗
  6. 6mo ago

    2.0.0-beta.21 drops Stability AI image generation

    Support for generating images through the Stability AI API is removed, narrowing image work to the Gemini path added days earlier. A fix for extracting images by URL in chat history accompanies it.

    View source ↗