2.0.0-beta.19

🎁 New

• Add Support for Gemini Pro 3
• Add Support for Minimax M2

🛠️ Fix

• Fix issue with new Agent safety checker
• Test khoj.el with newer Emacs versions

Full Changelog: 2.0.0-beta.18...2.0.0-beta.19

2.0.0-beta.18

🎁 New

• Support Exa as fast webpage reader
• Support using MCP tools in research mode by @debanjum in #1229

🧪 Improve

• Show agent creation validation error in UX
• Improve new agent safety checker
• Upgrade to Firecrawl v2 web search, webpage read providers

🛠️ Fix

• Drop support for Firecrawl webpage summarizer

Full Changelog: 2.0.0-beta.17...2.0.0-beta.18

2.0.0-beta.17

🎁 New

• Add support for Exa as web search provider

🧪 Improve

• Improve cost, usage tracking for Grok 4, Claude 4.5, Kimi K2 Thinking models

🛠️ Fix

• Drop support for Jina for web search, webpage read
• Remove deprecated /default slash command from UX slash command options

Full Changelog: 2.0.0-beta.16...2.0.0-beta.17

2.0.0-beta.16

🧪 Improve

• Evaluate Khoj on the same set of questions across modes in a single eval run
• Upgrade server, web and desktop package dependencies

🛠️ Fix

• Handle regression in using AI models via Groq API for research mode by @debanjum in 2fdb1fc

Full Changelog: 2.0.0-beta.15...2.0.0-beta.16

2.0.0-beta.15

🧪 Improve

• Create images with different aspect ratios (i.e portrait, landscape, square) via user instructions by @debanjum in 1e81b51
• Support Cerebras AI model API in dd8e805
• Allow Server Admins to set Chat Models for Fast vs Deep thinking in 4976b24

🛠️ Fix

• Fix indexing Notion knowledge bases in 00f0d23
• Only show tools that have been already configured to agents

Full Changelog: 2.0.0-beta.14...2.0.0-beta.15

2.0.0-beta.14

🧪 Improve

• Improve Khoj research tools and Openai api usage in 9f6aa92
• Improve table, train of thought rendering, show disconnect errors on web app in b99ccbc, 13d26ae and 59bfaf9
• Improve Gpt-oss, Deepseek V3.1 model support in 9f6aa92, 464c154
• Fallback to Groq Flex Service Tier for higher rate limits in 29ae476

🛠️ Fix

• Fix to create chat with custom agents from Obsidian by @Fh26697 in #1216
• Fix and improve file read, write handling in Obsidian by @debanjum in 3c3205b

New Contributors

• @Fh26697 made their first contribution in #1216

Full Changelog: 2.0.0-beta.13...2.0.0-beta.14

2.0.0-beta.13

🎁 New

• Add support for GPT 5 model series by @debanjum in 564adb2

🧪 Improve

• [Breaking Change] Drop native offline chat and server file indexing in #1212
• Improve researcher agent and its tools in 0e1615a, a79025e, 0a6d870, cbeefb7
• Improve cache and reasoning utilization for official openai model in b2d2608

⚙️ Developer

• Speed up app development using a faster, modern toolchain in #1196

Full Changelog: 2.0.0-beta.12...2.0.0-beta.13

2.0.0-beta.12

🎁 New

• Add support for Grok 4

🧪 Improve

• Improve extracting thoughts from reasoning models
• Improve web, semantic search tool context shown to research agent

Full Changelog: 2.0.0-beta.11...2.0.0-beta.12

2.0.0-beta.11

What's Changed

• Resolve Pydantic deprecation warnings by @emmanuel-ferdman in #1211

Full Changelog: 2.0.0-beta.10...2.0.0-beta.11

2.0.0-beta.9

🧪 Improve

• Ease bi-directional messaging on web app using websockets by @debanjum in #1207
  This should make it easier and faster to guide Khoj while it is doing research by sending messages from the chat page on web app

Full Changelog: 2.0.0-beta.8...2.0.0-beta.9