v1.12.1

What's changed

• Added LM Studio provider
• Fixed issues with function calling with OpenAI compatible providers.
• Fixed issues with JSON parsing and extraction.

Full Changelog: v1.12.0...v1.12.1

v1.12.0

What's Changed

This is a major release featuring a complete architectural overhaul. We've removed Langchain in favor of a custom implementation for low-level control over streaming, generation, and provider-specific features. This refactor lays the groundwork for new capabilities in upcoming releases.

General Updates

• Renamed providers to connections for better UX. Added switch config field, separated personalization and preferences.
• Fixed SearxNG build and startup errors.
• Added new agents DIR, migrated suggestions, video and audio agents.
• Added message and chunk types for better type safety.
• Added code block rendering with syntax highlighting.
• Added version number and GitHub link in settings.
• Various UI improvements with animations via Framer Motion, fixed theme inconsistencies, and updated numerous small but noticeable details. Overall the UI is more reactive now.

Langchain Removal & Custom Provider System

Removed Langchain and implemented custom base and provider classes with streamText, streamObject, generateText and generateObject methods. We've also implemented function calling with partial streaming so the app now uses it instead of XML parsing (was a chaotic one).

Langchain is removed to get low-level control over how things are streamed, how things are generated, etc. This allows us to add provider-specific features like structured outputs and fine-tune the generation process.

New Session Manager

Added a new session manager where each session has an ID and is recorded for 30 minutes. Sessions can be restored via the ID if your page refreshes or the stream disconnects. The manager includes methods for block streaming, replaying, multiple listeners, etc.

Automatic Stream Reconnection: If the page gets refreshed or the stream is disconnected, you'll now be able to automatically reconnect to the stream. The app resumes from where the LLM left off and streams back the entire response that was generated before, allowing messages to continue without waiting for completion.

Widgets

Added widgets that are extracted and executed in parallel to the research. They fetch specific types of data like weather, stock information, and calculated results (more widgets coming soon). Each widget is represented with a specific UI card which enhances the UX.

Widgets currently available:

• Weather information
• Stock market data
• Mathematical calculations

Query Classification & Researcher with Actions

Query Classification: Each query is now classified before research begins, ensuring proper sources and widget information extraction.

Researcher Class with Actions: The researcher uses actions (which use tool calling), along with specific actions for reasoning. This allows the researcher to have many actions like web search, scraping, specific website search, discussion search, file uploads search, etc. It enables the possibility of adding more actions and integrations like Google Workspace (Calendar, Gmail, etc.) in future releases.

Deep Research Mode (Beta)

Added quality search mode (deep research) where the LLM reasons iteratively over the collected data to predict the next action. It takes appropriate actions until it's confident it has enough information to answer, then prepares a long, comprehensive report. This allows deeper exploration of topics and preparing detailed reports.

Note: Deep research is currently in beta. There's a lot more to come in this feature like creating documents, images, files, charts, etc. to prepare proper reports.

Updated Research Modes

Balanced Mode: Now also forces the LLM to reason (though it only has a limited maximum number of actions for controlled token usage).

Speed Mode: Runs without any reasoning for fast responses.

Block & JSON Patch Based Streaming

Now each and every step of the LLM is streamed so you can see what the LLM is doing, so it doesn't feel stuck. This provides real-time visibility into the research process.

New Upload Manager & Enhanced File Search

Added a new upload manager and upload store for handling file uploads and searching. It supports PDF, DOCX, and text files with no hard limits on file size or number (within reasonable limits).

The manager allows searching across multiple files with queries that are then reranked and selected via RRF (Reciprocal Rank Fusion), enhancing retrieval quality significantly.

Redesigned Source System

Focus modes are removed and replaced with 3 sources: web, academic, and discussions. They allow searching for what they say, can be toggled individually, and you can use all 3 at once or disable external data sources entirely to work only with uploaded files.

---

Full Changelog: v1.11.2...v1.12.0

v1.11.2

1.11.2

What's changed

• Fixed issues with Transformer models not loading
• Fixed issues with model selection not updating the state
• Fixed issues with empty message being sent

Full Changelog: v1.11.1...v1.11.2

v1.11.1

v1.11.1

What's changed

• Fixed issues with SearxNG requests getting timed out
• Model select values are now sourced off localStorage
• Run commands are updated to use volumes for persistent storage

Full Changelog: v1.11.0...v1.11.1

v1.11.0

v1.11.0

What's Changed

Major Features

• New Setup Wizard: New initial configuration experience with model selection and provider setup
• Configuration Management Overhaul: Completely redesigned config system with live updates and migration support
• Enhanced UI/UX: New sidebar, improved theme system and responsive design updates across all components
• Settings Dialog: New settings interface with model selector, embedding configuration and system instructions
• Single command installation: Perplexica can be now be installed using a single Docker command without pulling the whole repo.

Provider & Model Updates

• Added support for latest AI models: GPT-5, GPT-5 Mini, O3, Claude Opus 4.1, and Gemini 2.5 models
• New providers: AIML API, Lemonade, LM Studio and Transformers (with dynamic loading)
• Enhanced provider management with unified model registry
• Improved Groq integration using @langchain/groq
• Models are now fetched dynamically instead of being hardcoded

UI Improvements

• Mobile settings button for better accessibility on smaller screens
• Weather widget with geolocation support and Imperial/Metric units
• Enhanced Discover page with topic selection and preview mode
• Improved file attachment modal with light theme compatibility
• Auto scroll in chat with smart scroll back detection

Developer Experience

• API route validation to prevent malformed requests
• Separate message entity handling for better performance
• Citation component with improved rendering
• Hash based config management for efficient updates
• Instrumentation based migrations replacing ts-node

Fixes

• Fixed repeated first token and think tag handling issues
• Resolved double stringify JSON parsing errors
• Fixed wrong history during message rewriting
• Corrected file upload errors with DOC files
• Fixed iOS input zoom issues
• Resolved light mode color issues across components

New Contributors

• @ClawCloud-Ron made their first contribution in #731
• @haddadrm made their first contribution in #743
• @alckasoc made their first contribution in #781
• @glitchySid made their first contribution in #801
• @koyasi777 made their first contribution in #785
• @Davixk made their first contribution in #784
• @D1m7asis made their first contribution in #815
• @kittrydge made their first contribution in #817
• @chriswritescode-dev made their first contribution in #787
• @amoshydra made their first contribution in #827
• @boarder2 made their first contribution in #830
• @tuxthepenguin84 made their first contribution in #832
• @skoved made their first contribution in #855
• @justinmayer made their first contribution in #828
• @agungbesti made their first contribution in #866
• @fizikiukas made their first contribution in #876
• @ramkrishna2910 made their first contribution in #877
• @ruturaj-rathod made their first contribution in #880

Full Changelog: v1.10.2...v1.11.0

v1.10.2

What's Changed

• Add system instructions to customize model output.
• Add Deepseek chat model provider
• Support citation format [1,2,3,4] in message box.
• Update Gemini embedding models

New Contributors

• @singleparadox made their first contribution in #719

Full Changelog: v1.10.1...v1.10.2

v1.10.1

What's Changed

• Fix ONNX issues causing transformer models to not work.
• Fix issues with Gemini & Anthropic models
• feat(models): Update Gemini chat models by @ottsch in #705
• feat(api): add streaming support to search route by @OmarElKadri in #703

New Contributors

• @ottsch made their first contribution in #705
• @OmarElKadri made their first contribution in #703

Full Changelog: v1.10.0...v1.10.1

v1.10.0

What's Changed

• Add keep_alive parameter for Ollama models to specify how long models stay in memory.
• Add file uploads, allowing answers to be generated based on uploaded files (PDF, TXT, and DOCX supported for now).
• Add Gemini chat model providers.
• Add meta search agent class to improve maintainability and facilitate focus mode creation.
• Fix UI bugs (wrong colors, content overflowing).
• Add new prompt for generating blog-style answers.
• Update chat model providers and include the latest models.
• Pause the currently playing video in lightbox mode when the user switches to another video.
• Fix bugs related to custom OpenAI handling.
• Ensure only messages belonging to a chat are deleted, instead of all messages.
• Add automatic image and video search functionality.
• Add new settings page with live config refresh.
• Scroll into view only when a user message is sent.
• Remove backend and move all functionality to Next.js routes.
• Add thinking model support.
• Use a standalone build to reduce build sizes (from 1.8GB and 1.6GB to ~220MB).

New Contributors

• @timoa made their first contribution in #466
• @yslinear made their first contribution in #519
• @bart-jaskulski made their first contribution in #523
• @sainadh-d made their first contribution in #548
• @realies made their first contribution in #555
• @hacking-racoon made their first contribution in #531
• @data5650 made their first contribution in #599
• @user1007017 made their first contribution in #631
• @QuietlyChan made their first contribution in #649
• @wellCh4n made their first contribution in #642
• @ShortCipher5 made their first contribution in #655
• @ericdachen made their first contribution in #663

Full Changelog: v1.9.3...v1.10.0

v1.10.0 RC

What's new

• Added file upload functionality in webSearch focus mode. Currently, if any other search mode is selected other than web search, it will automatically change it to web search (As implementing file uploads in other focus modes is left, it will be added by the meta search agent class which will be introduced in the next version)
• Added keep_alive variable to specify the time for how long the model should be loaded in Ollama (Defaults to '5m')

v1.9.3

What's new

• Automatic focus on opening the site (on mount).

Fixes

• markdown-to-jsx package has been updated to fix issues with unclosed code block causing client side errors.
• The sources for summarization now have a maximum limit of 15 to prevent context errors.
• Downgraded/fixed NodeJS Docker version to prevent build errors caused by mismatch in package files.