v0.8.4

What's Changed

Features

• Add ALLOW_INSECURE_COOKIES feature flag by @nsarrazin in #1076
• Use pino for logs by @nsarrazin in #1086
• Add the simplest healthcheck route by @nsarrazin in #1089
• Move vars to dynamic, add metrics by @nsarrazin in #1085
• [VertexAI] Add support for tools parameter by @ArthurGoupil in #1065

Bugfixes

• 🎨 Add color for copy icon by @damianpumar in #1070
• set sameSite to lax when allowing insecure cookies by @nsarrazin in #1078

Other changes

• Sort by Trending to default by @KingNishHF in #1082
• retry text area height by @gary149 in #1091
• Kubernetes infra by @rtrompier in #1088
• fix(chart): update service metrics port name by @rtrompier in #1101
• fix(github): docker deploy version by @rtrompier in #1102
• fix(github): docker deploy version by @rtrompier in #1103

New Contributors

• @damianpumar made their first contribution in #1070
• @gugarosa made their first contribution in #1081
• @KingNishHF made their first contribution in #1082
• @rtrompier made their first contribution in #1088
• @ArthurGoupil made their first contribution in #1065

Full Changelog: v0.8.3...v0.8.4

v0.8.3

What's Changed

Features

• add an endpoint to api to get an assistant by id by @alak in #1020
• Move featured assistant check behind REQUIRE_FEATURED_ASSISTANTS by @nsarrazin in #1043
• Support Gemini 1.5 Pro from Vertex AI by @flexchar in #1041

Bugfixes

• Allow websearch if assistant has dynamic prompts by @nsarrazin in #1017
• Add assistant instructions to first system message by @nsarrazin in #1023
• fix: See all your assistants regardless of usercount by @nsarrazin in #1053
• Only require featured assistants for other users' assistants by @nsarrazin in #1054
• Final fix to assistants display logic by @nsarrazin in #1056
• add createdAt and udpatedAt to new messages by @nsarrazin in #1062
• add default value for description and examples in the API route for models by @alak in #1063
• return only valid models in the API by @alak in #1064

Other changes

• Add logs to tokenizer fetch by @nsarrazin in #1030
• return empty if error on hfapi endpoint by @nsarrazin in #1038
• Use user in docker image for spaces by @nsarrazin in #1045
• Docker chown change by @nsarrazin in #1046
• Always dispose of pipeline for embeddings after use by @nsarrazin in #1048
• Bump version to 0.8.3 by @nsarrazin in #1068

Full Changelog: v0.8.2...v0.8.3

v0.8.2

What's Changed

• Deploy HuggingChat changes manually only by @nsarrazin in #993
• Release Gemma 7B 1.1 in HuggingChat by @nsarrazin in #994
• Add gemma 7B it to old models by @nsarrazin in #995
• Update models and add check for assistants model on startup by @nsarrazin in #998
• Add Command R+ to HuggingChat config by @nsarrazin in #1001
• Add prompt examples for command-r-plus by @nsarrazin in #1002
• modal wording by @gary149 in #1000
• Fix unresponsive page after deleting all conversations by @francoislanc in #1004
• Show the assistant model name in the chat input by @nsarrazin in #1006
• Send characters by groups of 5 by @nsarrazin in #1007
• Add new model to HuggingChat config by @nsarrazin in #1008
• Tweak default Zephyr example by @lewtun in #1010
• Fix capitalization by @everypizza1 in #1012
• Add langserve endpoint by @antonioramos1 in #1009
• Update marked from v4 to v12 by @Saghen in #1015
• Bump version to 0.8.2 by @nsarrazin in #1016

New Contributors

• @francoislanc made their first contribution in #1004
• @lewtun made their first contribution in #1010
• @everypizza1 made their first contribution in #1012
• @antonioramos1 made their first contribution in #1009

Full Changelog: v0.8.1...v0.8.2

v0.8.1

Mostly a small update to trigger the CI/CD release workflow 😄

Some of the big changes:

• You can now use tokenizer to set the chat formatting (it will apply the jinja template)
• We have Cohere endpoints in chat-ui now!
• The default chat template if you don't set one, or a tokenizer is now ChatML.

What's Changed

• Bump vite from 4.5.2 to 4.5.3 by @dependabot in #974
• Use jinja template for chat formatting (#730) by @nsarrazin in #744
• Add support for cohere endpoints by @nsarrazin in #976
• quick fix error message by @nsarrazin in #977
• Bump undici from 5.28.3 to 5.28.4 by @dependabot in #979
• Set ChatML as default chat prompt template by @nsarrazin in #985
• Update package.json version by @zacps in #983
• Move default template so it doesn't override tokenizer by @nsarrazin in #987
• Publish docker images based on package.json version by @nsarrazin in #988
• Make sure CI/CD is the same for no DB image by @nsarrazin in #989
• Use raw for versioning by @nsarrazin in #990
• Use proper semver for major minor & patch releases in CI/CD by @nsarrazin in #991
• Bump version to 0.8.1 by @nsarrazin in #992

Full Changelog: v0.8...v0.8.1

v0.8

What's Changed

• Make mixtral the default model for new users by @nsarrazin in #672
• Update share behaviour by @mishig25 in #645
• Use JSON5 for parsing all config vars by @nsarrazin in #671
• Fixes "invalid url" error in websearch by @muhammadbaqirjafari in #675
• Remove falcon 180B by @nsarrazin in #680
• Apply settings without reloading the page by @nsarrazin in #682
• Add doc section about updating HuggingChat by @nsarrazin in #683
• Add embedding models configurable, from both transformers.js and TEI by @mikelfried in #646
• Update openchat to 0106 by @nsarrazin in #687
• Mv embeddingEndpoints.ts by @mishig25 in #688
• Add an endpoints to expose models and conversations by @alak in #694
• Set message id on server side by @nsarrazin in #696
• Add NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO by @nsarrazin in #700
• [Small model] Explicitly close AsyncGenerator by @mishig25 in #704
• Bump max new tokens for NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO by @nsarrazin in #702
• Add conversation API endpoint to get conversation by @alak in #698
• Revert "[Small model] Explicitly close AsyncGenerator" by @mishig25 in #709
• Bump vite from 4.3.9 to 4.5.2 by @dependabot in #713
• [Websearch] Fix UI flicker by @mishig25 in #710
• [JS] use shorthand notation by @mishig25 in #711
• [Fix] UI bug "Delete" label by @mishig25 in #717
• Bring back mistral task model by @nsarrazin in #718
• Use JSON5 for embeddings by @nsarrazin in #719
• Continue generation feature by @nsarrazin in #707
• Update readme to use mistralai/Mistral-7B-Instruct-v0.2 by @mishig25 in #723
• [SmallModel] Nit update the comment by @mishig25 in #726
• close the stream after the final answer by @alak in #722
• [Fix] small fixes in websearch generateQuery by @mishig25 in #727
• Assistants feature by @nsarrazin in #639
• fix: social thumbnail assistants in prod by @nsarrazin in #732
• Assistants: quick layout fixes by @gary149 in #733
• [SmallModel] TASK_MODEL='mistralai/Mistral-7B-Instruct-v0.1' by @mishig25 in #729
• Fix social thumbnails: images by @nsarrazin in #734
• Add failure strategy to README.md by @nsarrazin in #731
• move assistants settings buttons by @gary149 in #738
• Update nous prompt by @nsarrazin in #728
• switching api to public by @alak in #739
• [Fix README] Nit: some small indent issues by @mishig25 in #741
• Add CodeLlama-70b-Instruct-hf by @nsarrazin in #752
• Update news for code llama by @mishig25 in #753
• Web search: UI update by @gary149 in #749
• Remove tabs in websearch prompt by @nsarrazin in #754
• Mobile navs: bigger buttons and hide new? by @gary149 in #720
• Top assistants page by @nsarrazin in #740
• assistants: add beta tag by @gary149 in #758
• Hide websearch on assistants by @nsarrazin in #759
• Assistants: UI tweaks for release by @gary149 in #764
• add an api endpoint to get current user's informations by @alak in #763
• Bring back featured check on top assistants page by @nsarrazin in #765
• Convert all assistants avatar to jpeg server-side by @nsarrazin in #762
• Make assistant feature visible globally by @nsarrazin in #756
• assistants: add hint by @gary149 in #767
• Use assistant avatar in conversation, if available by @nsarrazin in #776
• Disable websearch on assistants, server & client side by @nsarrazin in #778
• [Assistants] Add pagination by @mishig25 in #772
• Add Plausible analytics by @nsarrazin in #779
• add alternative callback url for the login process by @alak in #766
• [Assistnats] Fix pagination numTotalPages by @mishig25 in #780
• [Assistants] Fix/clear searchParams while changing models by @mishig25 in #783
• [Assistants] Fix selected model reactivity by @mishig25 in #784
• [Fix] on cancel operations, preerve previous URL query params by @mishig25 in #781
• Several QoL contributions by @flexchar in #760
• [Chat input] Fix asian keyboards by @mishig25 in #789
• Update README.md by @eltociear in #791
• [Login Modal] Fix not showing properly on mobile by @mishig25 in #796
• [Assistnats] improve reporting by @mishig25 in #797
• Add missing prod config for report assistants by @mishig25 in #799
• [Assitants] Community | User tabs by @mishig25 in #773
• [Frontend] keypress -> keydown by @mishig25 in #800
• [Chat input] Blur on submit by @mishig25 in #790
• [Assistants] update page & visiblity logic by @mishig25 in #802
• fix continue button on openai endpoint by @nsarrazin in #803
• [Privacy page] Skip disclaimer modal by @mishig25 in #808
• /assistants: add user count by @gary149 in #809
• [Assistants] update numbering labels by @mishig25 in #811
• Fix delete assistant query performance by @nsarrazin in #806
• [fix] If APP_BASE is empty, go to '/' instead of an empty string. by @gururise in #785
• Only use plausible if PUBLIC_ORIGIN is set by @nsarrazin in #814
• 🗃️ Create index on users.username by @coyotte508 in #819
• Added support for SearXNG as a websearch source by @Karlostavitch1 in #805
• Make lorax openai endpoint working by @flozi00 in #755
• Simple Email Authorization by @nicomt in #801
• Ask for a reason when reporting assistants by @nsarrazin in #825
• [Assistants] Add new indices for efficient querying by @mishig25 in #810
• [Assisntats reports] Minor fixes by @mishig25 in #827
• Conversation trees (#223) by @nsarrazin in #807
• [Privacy page] Skip disclaimer modal including basePath by @mishig25 in #829
• Allow customising defaultQuery for native Azure OpenAI support by @zacps in #822
• [Assistants settings] Fix freezing by @mishig25 in #830
• Revert "[Assistants settings] Fix freezing" by @mishig25 in #831
• [Mongo] Optimize reports collection query by @mishig25 in #832
• Revert "[Mongo] Optimize reports collection query" by @mishig25 in #833
• [Mongo] Optimize reports collection query (v2) by @mishig25 in #834
• fix shifted layout by @gary149 in #836
• Make compatible with Microsoft Entra (v1 and v2 token...

v0.7

Pretty big changelog, as it's been a while since we did a release. 😁 Some of the most important changes:

• We now support a whole bunch of different backends. (TGI, SageMaker, llama.cpp and all OpenAI compatible backends)
• We added support for more search providers (we now support SerpApi, serper.dev, SerpStack, You.com, and no search provider at all by parsing search results directly!)
• We added support for multi-sessions for users
• We added a new redesigned settings menu!
• We added new chat-ui and chat-ui-db docker images that you can run directly from GHCR
• And a lot of QoL changes and bugfix, feel free to read the changelog for more.

I'll split up future releases into smaller chunks to make it easier to parse the changes, this was the big one!

What's Changed

• Update README.md by @gary149 in #435
• Mobile: fix model selection by @gary149 in #448
• Add latex support with marked-katex-extension by @julien-blanchon in #450
• Bump mongodb from 5.3.0 to 5.8.0 by @dependabot in #424
• Improve error message when the .env MODELS is not well formatted by @julien-blanchon in #451
• Update embedding model for WebSearch by @mishig25 in #437
• Backend refactor for streaming endpoints by @nsarrazin in #444
• Update README.md by @mishig25 in #455
• Refactor summarization so it gets called from backend by @nsarrazin in #456
• Make embedding model settings more future-proof by @mishig25 in #454
• fix JSON.parse for summerize by @jackielii in #445
• Add a message wide copy button by @julien-blanchon in #453
• Add manifest.json by @nsarrazin in #469
• Privacy update & readme linting by @nsarrazin in #472
• Readme: Update No Setup by @gary149 in #473
• Store preprompt in conversation (#422) by @nsarrazin in #470
• Remove shared routes by @nsarrazin in #478
• Move some tasks to small model by @nsarrazin in #479
• Bump postcss from 8.4.23 to 8.4.31 by @dependabot in #483
• Bump zod from 3.21.4 to 3.22.3 by @dependabot in #487
• Version upgrade to Svelte 4 by @nsarrazin in #488
• fix #475 by @gary149 in #490
• Add prompts templates to a markdown doc by @nsarrazin in #489
• Pass stop sequences properly to TGI by @nsarrazin in #499
• Add a job in release action that updates the .env in prod by @nsarrazin in #493
• Save user message on request (#337) by @nsarrazin in #492
• Added optional llm summerization by @itaybar in #498
• Login button sends you directly to login page by @nsarrazin in #501
• Feature customizable description by @julien-blanchon in #507
• Update websearch prompting & summary prompting by @nsarrazin in #503
• Feature/add OIDC optional tolerance and resource by @itaybar in #496
• Update CI scripts for dotenv management by @nsarrazin in #508
• Add zephyr to config by @nsarrazin in #510
• Title update occur when available by @nsarrazin in #512
• Switch chat model back to mistral, use zephyr for small tasks by @nsarrazin in #515
• Manifest update by @gary149 in #513
• Remove update banner for new websearch by @nsarrazin in #517
• Add logo by @gary149 in #518
• manifest: /chat/ by @gary149 in #522
• [Websearch] UI fix sources by @mishig25 in #521
• Break up SECRET_CONFIG into individual secrets by @nsarrazin in #523
• Misc julien tweaks by @julien-c in #525
• Bump undici and @sveltejs/kit by @dependabot in #524
• Bring back disclaimer modal & updates to guest mode by @nsarrazin in #526
• Add a more visible retry button on failure to generate by @nsarrazin in #530
• Add checkbox for displaying emojis in sidebar by @nsarrazin in #520
• add you.com integration by @sam-h-bean in #514
• [DOCS] Minor fixes in README.md by @Akash190104 in #532
• Show model modal even if only one model is available by @nsarrazin in #536
• Modular backends & support for openAI & AWS endpoints by @nsarrazin in #541
• Reset app title when clicking new chat by @nsarrazin in #552
• [websearch] Enable lazy construction of feature-extraction pipeline by @xenova in #554
• Add support for tgi multimodal models by @nsarrazin in #531
• Update Node.js version to 20 in CI by @nsarrazin in #564
• Option for running websearch locally by @nsarrazin in #563
• Disable cache on tgi endpoint by @nsarrazin in #567
• Add a CI check for dockerfile build by @nsarrazin in #557
• Bump sharp from 0.32.5 to 0.32.6 by @dependabot in #558
• Add ollama endpoint support by @nsarrazin in #569
• Add openchat to huggingchat config by @nsarrazin in #573
• [#547] fix stream generation bug by @jason9693 in #561
• fix: new chat when old one is still streaming by @ftgomez in #578
• Make chat title summarization non blocking by @nsarrazin in #575
• Open in new tab if cookies are disabled by @nsarrazin in #580
• Fix conversation preprompt by @galen-ft in #595
• 🔒️ Harden session ID generation by @coyotte508 in #599
• bump svelte & related to latest by @nsarrazin in #600
• Update dockerfile to node 20 by @nsarrazin in #601
• Only refresh cookie on POST by @nsarrazin in #606
• Session management improvements: Multi sessions, renew on login/logout by @nsarrazin in #603
• Add support for passing an API key or any other custom token in the authorization header by @galen-ft in #579
• Support websearch for openAI chat completion mode by @nsarrazin in #608
• Group conversations by date + small nav fixes by @nsarrazin in #582
• Get rid of the unused websearch collection, and the unused endpoint by @nsarrazin in #607
• Standardize HF_ACCESS_TOKEN -> HF_TOKEN by @Wauplin in #610
• Reworked settings menu by @nsarrazin in #591
• Misc UI by @gary149 in #616
• Add chat-ui and chat-ui-db docker images by @nsarrazin in #613
• Add mixtral to models list in huggingchat by @nsarrazin in #619
• Fix bugs related to preprompt & query params model setting by @nsarrazin in #621
• Add option to unlist a model by @nsarrazin in #625
• Conditionally show the data sharing toggle by @nsarrazin in #624
• Settings UI update by @gary149 in #638
• Add mistral 7b v0.2 by @nsarrazin in #626
• share model url: fix by @gary149 in #640
• Reduce mistral v0.2 temp by @nsarrazin in #643
• Add negating pattern to not ignore .env.local when running docker build by @gterziysky in #642
• Add allow & blocklist for websearch by @nsarrazin in https...

v0.6 Websearch update

What's Changed

• Fix auth on websearch by @nsarrazin in #410
• Add rate limiting to websearch and title summary by @nsarrazin in #433
• [Websearch] update by @mishig25 in #427
• Bump to 0.6.0 by @nsarrazin in #434

Full Changelog: v0.5...v0.6

v0.5 Amazon SageMaker & custom system prompts

We now support Amazon SageMaker as a backend! 🥳 Check out the updated README for more info.

We also added support for custom system prompts. This will let you customize the responses you expect from your assistants. You can try it out over at hf.co/chat !

A big thanks to our new contributors, and especially to @AndreasMadsen who created some great PRs in this release, including adding handlebars support so you can use your own custom prompt structure more easily.

What's Changed

• Added access token note by @merveenoyan in #360
• Update /privacy and other content following Llama v2 release by @julien-c in #374
• Clarify that model 'tokens' are not actual tokens by @AndreasMadsen in #367
• Attempt to clarify how hosted API ≠ local endpoint by @julien-c in #373
• Make model branding customizable based on env var by @flozi00 in #345
• trim and remove stop-suffixes from summary by @AndreasMadsen in #369
• Add a login button by @nsarrazin in #381
• allow different user and assistant end-token by @AndreasMadsen in #375
• Leverage model link to modelUrl when informed by @airibarne in #385
• feat: add support for endpoints requiring client authentication using PKI by @cambriancoder in #393
• Add Sagemaker support by @nsarrazin in #401
• fix docs regarding TGI endpoint url by @AndreasMadsen in #408
• Bump @antfu/utils from 0.7.2 to 0.7.6 by @dependabot in #407
• Bump word-wrap from 1.2.3 to 1.2.5 by @dependabot in #406
• Bump tough-cookie from 4.1.2 to 4.1.3 by @dependabot in #405
• Bump vite from 4.3.5 to 4.3.9 by @dependabot in #404
• Make all prompt templates configurable by @AndreasMadsen in #400
• Support custom system prompts from the user by @nsarrazin in #399

New Contributors

• @AndreasMadsen made their first contribution in #367
• @flozi00 made their first contribution in #345
• @airibarne made their first contribution in #385
• @cambriancoder made their first contribution in #393
• @dependabot made their first contribution in #407

Full Changelog: v0.4...v0.5

v0.4 Llama-2 on hf.co/chat !

Releasing v0.4 to celebrate the release of Llama-2 on HuggingChat! 🚀

What's Changed

• Basic rate limiter by @nsarrazin in #320
• Add support for websearch retries by @nsarrazin in #318
• Factor out HF_API_ROOT to allow different inference endpoints by @DayOfThePenguin in #332
• Add support for HF summarization endpoint in the websearch by @nsarrazin in #319
• Add optional timestamps to messages by @nsarrazin in #294
• Add ability to define custom model/dataset URLs by @secondtruth in #347
• Added Docker Space deployment docs by @merveenoyan in #350
• bump version to 0.4 by @nsarrazin in #353
• Minor improvement to headers in README by @merveenoyan in #354
• Option to disable login on first N messages by @nsarrazin in #352
• Support rate limit based on user IP by @nsarrazin in #342
• Reopen: Feature/disable login for n messages by @gary149 in #356
• Fix ToC + move up space template by @osanseviero in #359

New Contributors

• @DayOfThePenguin made their first contribution in #332
• @secondtruth made their first contribution in #347
• @merveenoyan made their first contribution in #350

Full Changelog: v0.3...v0.4

ChatUI v0.3

We're going to start doing tagged releases of ChatUI now, in order to easily generate changelogs in the future.

We're currently on version 0.3 but as it is the first release on GitHub, it will contain all the changes & contributors so far! 👋

New Contributors

• @julien-c made their first contribution in #1
• @Grsmto made their first contribution in #2
• @gary149 made their first contribution in #4
• @coyotte508 made their first contribution in #13
• @Pierrci made their first contribution in #110
• @osanseviero made their first contribution in #168
• @loubnabnl made their first contribution in #184
• @nsarrazin made their first contribution in #188
• @symsmith made their first contribution in #206
• @TP-O made their first contribution in #209
• @digitaljerrygit made their first contribution in #205
• @SteveSimms made their first contribution in #222
• @fredguth made their first contribution in #250
• @XciD made their first contribution in #259
• @CarolynM8ri3 made their first contribution in #279
• @eltociear made their first contribution in #286
• @CAPTAIN320 made their first contribution in #291
• @averad made their first contribution in #288
• @arifszn made their first contribution in #300
• @rogerserper made their first contribution in #302

Full Changelog: https://github.com/huggingface/chat-ui/commits/v0.3