v3.3.0

What's New

• UI Improvements: The minimum window size now adapts to the font size. A few labels and links have been fixed. The Embeddings Device selection of "Auto"/"Application default" works again. The window icon is now set on Linux. The antenna icon now displays when the API server is listening.
• Single Instance: Only one instance of GPT4All can be opened at a time. This is now enforced.
• Greedy Sampling: Set temperature to zero to enable greedy sampling.
• API Server Changes: The built-in API server now responds correctly to both legacy completions, and chats with message history. Also, it now uses the system prompt configured in the UI.
• Translation Improvements: The Italian, Romanian, and Traditional Chinese translations have been updated.

Contributors

• Jared Van Bortel (Nomic AI)
• Adam Treat (Nomic AI)
• 3Simplex (@3Simplex)
• Riccardo Giovanetti (@Harvester62)
• Victor Emanuel (@SINAPSA-IC)
• Dominik (@cosmic-snow)
• Shiranui (@supersonictw)

Full Changelog: CHANGELOG.md

v3.2.1

Fixed

• Fix a potential Vulkan crash on application exit on some Linux systems (#2843)
• Fix a bad CUDA build option that led to gibberish on newer NVIDIA GPUs (#2855)

Full Changelog: v3.2.0...v3.2.1

v3.2.0

Added

• Add Qwen2-1.5B-Instruct to models3.json (by @ThiloteE in #2759)
• Enable translation feature for seven languages: English, Spanish, Italian, Portuguese, Chinese Simplified, Chinese Traditional, Romanian (#2830)

Changed

• Add missing entries to Italian transltation (by @Harvester62 in #2783)
• Use llama_kv_cache ops to shift context faster (#2781)
• Don't stop generating at end of context (#2781)

Fixed

• Case-insensitive LocalDocs source icon detection (by @cosmic-snow in #2761)
• Fix comparison of pre- and post-release versions for update check and models3.json (#2762, #2772)
• Fix several backend issues (#2778)
  • Restore leading space removal logic that was incorrectly removed in #2694
  • CUDA: Cherry-pick llama.cpp DMMV cols requirement fix that caused a crash with long conversations since #2694
• Make reverse prompt detection work more reliably and prevent it from breaking output (#2781)
• Disallow context shift for chat name and follow-up generation to prevent bugs (#2781)
• Explicitly target macOS 12.6 in CI to fix Metal compatibility on older macOS (#2846)

New Contributors

• @SINAPSA-IC made their first contribution in #2828

Full Changelog: CHANGELOG.md

v3.1.1-web_search_beta_2

This is version 2 of the web search beta which contains some important fixes including upstream llama.cpp fixes for Llama 3.1.

Fixes

• Update to latest llama.cpp which includes RoPE fix
• Fix problem with only displaying one source for tool call excerpts
• Add the extra snippets to the source excerpts
• Fix the way we're injecting the context back into the model for web search
• Change the suggestion mode to turn on for tool calls by default

WARNING:

There was a problem with the synchronization between this beta release and the models.json. In order to make this work you have to perform the following steps:

1. Rename the file Meta-Llama-3.1-8B-Instruct-128k-Q4_0.gguf to Meta-Llama-3.1-8B-Instruct-Q4_0.gguf
2. Copy the following into the prompt template:

<|start_header_id|>user<|end_header_id|>

%1<|eot_id|><|start_header_id|>assistant<|end_header_id|>

%2

3. Copy the following into the system prompt:

<|start_header_id|>system<|end_header_id|>
Environment: ipython
Tools: brave_search

Cutting Knowledge Date: December 2023
Today Date: 25 Jul 2024

You are a helpful assistant.<|eot_id|> 

v3.1.1

Added

• Ability to add OpenAI compatible remote models (#2683)

Fixed

• Update llama.cpp to cherry-pick Llama 3.1 RoPE fix. (#2758)

v3.1.0-web_search_beta

This is the beta version of GPT4All including a new web search feature powered by Llama 3.1. To use this version you should consult the guide located here: https://github.com/nomic-ai/gpt4all/wiki/Web-Search-Beta-Release

For questions and feedback please join the discord channel: https://discord.com/invite/4M2QFmTt2k

v3.1.0

Added

• Generate suggested follow-up questions (#2634)
• Scaffolding for translations (#2612)
• Spanish (MX) translation (by @jstayco in #2654)
• Chinese (Simplified) translation by mikage (#2657)
• Dynamic changes of language and locale at runtime (#2659, #2677)
• Romanian translation by @SINAPSA_IC (#2662)
• Chinese (Traditional) translation (by @supersonictw in #2661)
• Italian translation (by @Harvester62 in #2700)

Changed

• Customize combo boxes and context menus to fit the new style (#2535)
• Improve view bar scaling and Model Settings layout (#2520
• Make the logo spin while the model is generating (#2557)
• Server: Reply to wrong GET/POST method with HTTP 405 instead of 404 (by @cosmic-snow in #2615)
• Update theme for menus (by @3Simplex in #2578)
• Move the "stop" button to the message box (#2561)
• Build with CUDA 11.8 for better compatibility (#2639)
• Make links in latest news section clickable (#2643)
• Support translation of settings choices (#2667, #2690)
• Improve LocalDocs view's error message (by @cosmic-snow in #2679)
• Ignore case of LocalDocs file extensions (#2642, #2684)
• Update llama.cpp to commit 87e397d00 from July 19th (#2694)
  • Add support for GPT-NeoX, Gemma 2, OpenELM, ChatGLM, and Jais architectures (all with Vulkan support)
  • Enable Vulkan support for StarCoder2, XVERSE, Command R, and OLMo
• Show scrollbar in chat collections list as needed (by @cosmic-snow in #2691)

Removed

• Remove support for GPT-J models (#2676, #2693)

Fixed

• Fix placement of thumbs-down and datalake opt-in dialogs (#2540)
• Select the correct folder with the Linux fallback folder dialog (#2541)
• Fix clone button sometimes producing blank model info (#2545)
• Fix jerky chat view scrolling (#2555)
• Fix "reload" showing for chats with missing models (#2520
• Fix property binding loop warning (#2601)
• Fix UI hang with certain chat view content (#2543)
• Fix crash when Kompute falls back to CPU (#2640)
• Fix several Vulkan resource management issues (#2694)

v3.0.0

What's New

• Complete UI overhaul (#2396)
• LocalDocs improvements (#2396)
  • Use nomic-embed-text-v1.5 as local model instead of SBert
  • Ship local model with application instead of downloading afterwards
  • Store embeddings flat in SQLite DB instead of in hnswlib index
  • Do exact KNN search with usearch instead of approximate KNN search with hnswlib
• Markdown support (#2476)
• Support CUDA/Metal device option for embeddings (#2477)

Fixes

• Fix embedding tokenization after #2310 (#2381)
• Fix a crash when loading certain models with "code" in their name (#2382)
• Fix an embedding crash with large chunk sizes after #2310 (#2383)
• Fix inability to load models with non-ASCII path on Windows (#2388)
• CUDA: Do not show non-fatal DLL errors on Windows (#2389)
• LocalDocs fixes (#2396)
  • Always use requested number of snippets even if there are better matches in unselected collections
  • Check for deleted files on startup
• CUDA: Fix PTX errors with some GPT4All builds (#2421)
• Fix blank device in UI after model switch and improve usage stats (#2409)
• Use CPU instead of CUDA backend when GPU loading fails the first time (ngl=0 is not enough) (#2477)
• Fix crash when sending a message greater than n_ctx tokens after #1970 (#2498)

New Contributors

• @woheller69 made their first contribution in (#2339)
• @patcher9 made their first contribution in (#2386)
• @sunsided made their first contribution in (#2414)
• @johnwparent made their first contribution in (#2319)
• @mcembalest made their first contribution in (#2488)

Full Changelog: v2.8.0...v3.0.0

v2.8.0

What's New

• Context Menu: Replace "Select All" on message with "Copy Message" (#2324)
• Context Menu: Hide Copy/Cut when nothing is selected (#2324)
• Improve speed of context switch after quickly switching between several chats (#2343)
• New Chat: Always switch to the new chat when the button is clicked (#2330)
• New Chat: Always scroll to the top of the list when the button is clicked (#2330)
• Update to latest llama.cpp as of May 9, 2024 (#2310)
• Add support for the llama.cpp CUDA backend (#2310, #2357)
  • Nomic Vulkan is still used by default, but CUDA devices can now be selected in Settings
  • When in use: Greatly improved prompt processing and generation speed on some devices
  • When in use: GPU support for Q5_0, Q5_1, Q8_0, K-quants, I-quants, and Mixtral
• Add support for InternLM models (#2310)

Fixes

• Do not allow sending a message while the LLM is responding (#2323)
• Fix poor quality of generated chat titles with many models (#2322)
• Set the window icon correctly on Windows (#2321)
• Fix a few memory leaks (#2328, #2348, #2310)
• Do not crash if a model file has no architecture key (#2346)
• Fix several instances of model loading progress displaying incorrectly (#2337, #2343)
• New Chat: Fix the new chat being scrolled above the top of the list on startup (#2330)
• macOS: Show a "Metal" device option, and actually use the CPU when "CPU" is selected (#2310)
• Remove unsupported Mamba, Persimmon, and PLaMo models from the whitelist (#2310)
• Fix GPT4All.desktop being created by offline installers on macOS (#2361)

Full Changelog: v2.7.5...v2.8.0

v2.8.0-pre1

What's New

• Context Menu: Replace "Select All" on message with "Copy Message" (#2324)
• Context Menu: Hide Copy/Cut when nothing is selected (#2324)
• Improve speed of context switch after quickly switching between several chats (#2343)
• New Chat: Always switch to the new chat when the button is clicked (#2330)
• New Chat: Always scroll to the top of the list when the button is clicked (#2330)
• Update to latest llama.cpp as of May 9, 2024 (#2310)
• Add support for the llama.cpp CUDA backend (#2310)
  • Nomic Vulkan is still used by default, but CUDA devices can now be selected in Settings
  • When in use: Greatly improved prompt processing and generation speed on some devices
  • When in use: GPU support for Q5_0, Q5_1, Q8_0, K-quants, I-quants, and Mixtral
• Add support for InternLM models (#2310)

Fixes

• Do not allow sending a message while the LLM is responding (#2323)
• Fix poor quality of generated chat titles with many models (#2322)
• Set the window icon correctly on Windows (#2321)
• Fix a few memory leaks (#2328, #2348, #2310)
• Do not crash if a model file has no architecture key (#2346)
• Fix several instances of model loading progress displaying incorrectly (#2337, #2343)
• New Chat: Fix the new chat being scrolled above the top of the list on startup (#2330)
• macOS: Show a "Metal" device option, and actually use the CPU when "CPU" is selected (#2310)
• Remove unsupported Mamba, Persimmon, and PLaMo models from the whitelist (#2310)

Full Changelog: v2.7.5...v2.8.0-pre1