Patch Notes

Fix display quota problem display on /profile

Fix model availability status checking, preventing false "unavailable" status when models are operational

New Suggestions system to help users find relevant prompts and models

Added Onboarding flow for new users

Refactored Terminal architecture for better stability and performance (Context API)

Optimized Full-Stack mode UI - "Full Stack Mode" button is now hidden when active

Security hardening and improved usage quota management

SEO improvements and translation updates

Fix problem on search history

File Explorer (Full-stack mode) context menu - Right-click to create files, folders, or delete items (Windows-style)

Add File/Folder buttons in File Explorer (Full-stack mode) header with inline input for quick creation

Empty folders are now visible in the File Explorer (Full-stack mode)

New landing page

Clicking on a folder now selects it (highlighted) in addition to expanding/collapsing (Full-stack mode)

Add more providers to Image Generation Studio - now support OpenAI and Requesty in addition to ZAI

Enhanced image generation settings with more parameters and options

Fix api token creation when authentication from Google

Fix mobile navigation and translation display

Fix tool execution flow for OpenRouter models

Add Russian translation support - full i18n coverage for en, es, fr, ru

Add new mode for Generation Image models (support glm-image & cogview)

Major style redesign - standardized color scheme across all pages to cohesive blue theme

Add a new feature to make an app builder in real time

Removed highlight.js and react-syntax-highlighter dependencies (~1.1MB bundle size reduction) - using custom ReactSyntaxHighlighter with Z.ai theme

Improved LaTeX rendering with code block protection - prevents regex conflicts between math expressions and code

Fixed Mermaid diagram rendering conditions for better reliability

Add new GPT-5.2 series models (gpt-5.2, gpt-5.2-pro, gpt-5.2-chat-latest)

Add gpt-5-chat-latest and gpt-5.1-codex-max models

Add gpt-image-1.5 and chatgpt-image-latest models with all quality/size variants

Add Glm-4.7 models

Update o1 and o1-mini context length to 200,000 tokens

Update OpenAI embeddings pricing (text-embedding-3-small: $0.02, text-embedding-3-large: $0.13)

Remove deprecated OpenAI models (codex-mini-latest, text-moderation-latest)

Fix security vulnerability in libs

Add support to Gemini 3.0 models

Add a new function to filter models by capabilities

New Artifact - Full-Stack IDE: A complete in-browser development environment supporting multiple files and live preview

Improve artifact renderer performance and stability

Update OpenAI models database with latest pricing and models from official OpenAI pricing page

Add new GPT-5.1 series models (gpt-5.1, gpt-5.1-chat-latest, gpt-5.1-codex, gpt-5.1-codex-mini)

Add GPT-5 Pro and GPT-5 Search API models

Add new GPT-4o dated model (gpt-4o-2024-05-13)

Add new audio and realtime models (gpt-audio, gpt-audio-mini, gpt-realtime, gpt-realtime-mini)

Add GPT Image 1 Mini models with all quality/size variants

Add Sora 2 video generation models (sora-2, sora-2-pro, sora-2-pro-hd)

Add transcription diarization model (gpt-4o-transcribe-diarize)

Update all OpenAI model pricing to match current official rates

Better manage error from model instead of display generic msg

tracking usage on costflow for api/chat api calls

fix minor bugs

Add badge for display reasoning & others informatons on model list

Add model for glm-4.5v to coding plan

remove glm-4.6-air

Add Glm-4.6 models

Add Sonnet 4.5 models

Add Deepinfra provider

Add api key usage from Zai or Zai Coding Plan

Apikey not be retrieve from profile page

Google & Gemini Cli & Qwen Cli work correctly now !

Some OpenAi model have been fix !

Fix global vision capabilities

Add 7 news model from Z.ai and support for this new Provider

Improve Mistral & Codestral for support thinking mode

Translation for English, French and Spanish support (on it...)

Anthropic now works perfectly

Fix document management embedding but error on summary function (on fix...)

Add new supported providers (aiml, anthropic, cerebras, chutes, claude_code, codestral, deepinfra, deepseek, doubao, featherless, fireworks, google, google_cli, glama, groq, huggingface, io_intelligence, lite_llm, lm_studio, mistral, moonshot, ollama, openai, openrouter, perplexity, qwen_code, reka, requesty, together, xai) !

Add Byok mode for all supported providers (aiml, anthropic, cerebras, chutes, claude_code, codestral, deepinfra, deepseek, doubao, featherless, fireworks, google, gemini_cli, glama, groq, huggingface, io_intelligence, lite_llm, lm_studio, mistral, moonshot, ollama, openai, openrouter, perplexity, qwen_code, reka, requesty, together, xai), you can configure it on /profile page !

Add a research tool for model

Rework on all architecture for call model

Add a cache management for model list

Better management of cache for API

API mode bugfix who make return 400 error

Add new provider Requesty

Free Google Pro & Flash usage this weekend !

google/gemini-2.5-flash-image-preview can now generate image

Add Anthropic support provider

Fixed image prompt where AI was not correctly receiving the associated image

Fixed instability to send messages with some providers

Improve the error message when you have to log in (quota reached)

Assistant reasoning always says view instead of hide/view

Mermaid diagrams did not take into account the available width if we hid the reasoning ia

Can use your own api key for premium models

Due to abuse premium models need now a subscription or your own key

Add Grok 4 & Imagen 4

Improvements and bug fixes

Add mermaid support

Switch external api to /api/v1 instead of /api and update process in api-guide

Add copy button to code blocks

Better manage math rendering & code blocks

Add reprompt button to user message

Add delete button to all messages

SQL and Katex support conflict

No more create account without verification email

Free API plan is now limited to 5 requests and 1M tokens

Add Starter plan (2$/month) with 60 requests and 4.5M tokens

Fix streaming problem (reasoning & content)

Many security improvements

Add GPT-5 models support (gpt-5, gpt-5-mini, gpt-5-nano) with updated pricing and 400k context window

Update OpenAI models data with latest pricing information and cached input support

Fix retry & other functions

Increase global stability & performance

Fix authentication issue

Refactor the project to use the new architecture

Indicator of model status (operational, degraded, down, unknown)

Web search button (ask Qwen for What's the news on July 05, 2025?)

Fix many bugs (stream, model selector, loading,etc.)

Support Google Gemini native (beta)

Fix many bugs

Live search like ask Qwen for What's the news on June 19, 2025? 🤔 View assistant reasoning Here is the information available regarding June 19, 2025, according to the contextual sources provided: 1. Israel-Iran Conflict and European Defense 2. Cholera epidemic in the Democratic Republic of Congo (DRC) this date.

Complete Artifacts System - Create, execute, and share interactive code templates

React & Vue.js Support - Live component rendering with hooks and reactive data

Python Execution Environment - Script execution with virtual filesystem for file operations

External Templates Integration - Browse and use templates from GitHub, CodePen, and NPM

Intelligent Template Filtering - Automatically exclude incompatible frameworks (Docsify, GitBook)

Real-time Code Editor - Syntax highlighting with live preview and console logging

Add Support to MCP server for improve context

Add API mode to use for external APPs

Manage history conversation (ex: what is my two last questions ?)

Graphic redesign and standardization

Add API mode to use for external APPs

Manage history conversation (ex: what is my two last questions ?)

Fix issue with create new chat

Fixed an issue that prevented the xai model list from being retrieved correctly

Fixed an issue that prevented reasoning from displaying correctly

Add system prompt

Add system prompt statistics

Fix close button for search form

Add support to document summary

Better manage big documents with rag

Add 15 new models from chatgpt like: gpt-3.5-turbo-0125, gpt-4.1, gpt-4.1-mini, gpt-4.1-nano, gpt-4o, gpt-4o-mini, o1-mini, o3-mini, o4-mini

Add support to docx, pdf up to 30mo

Better manage rag context

Rename manually a chat

Add AI Dashboard to explore and compare the most powerful and innovative artificial intelligence models.

Add two new models: Grok 2 Image & mini (X.ai)

You can now generate images with the new model: Grok 2 Image (X.ai)

Add search functionality to find specific chats or messages

Rag can now use documents and/or messages from history

Better speed and stability

Stream issue fixed

Advanced settings are now dynamically updated based on the selected model

Fixed a login/logout issue that did not refresh history correctly

Fixed a bug where memory system was not working properly

Added the ability to ask questions about images while benefiting from contextual knowledge

Enhanced understanding of queries combining images and text for more relevant answers

Automatic selection of the best model capable of processing images when you attach one to your message

Fixed the issue "I can't see or analyze images" that could appear when using images

Ability to analyze PDF documents while benefiting from knowledge base for enriched answers

Simplified the interface for adding images and documents to conversations

Added Memory feature to remember user preferences and information across conversations

Enhanced user guide with dedicated sections for RAG and Memory features

Improved sitemap for better SEO and navigation

Fixed image compatibility warning to only show when relevant

Added local embeddings generation using Xenova/transformers

Implemented RAG (Retrieval Augmented Generation) for enhanced context awareness

Created reusable Tailwind components for consistent UI across the application

Optimized database queries for better performance

Fixed embedding generation for existing messages

Set dark theme as default with flash prevention during page load

Edit and retry functionality don't present any bug anymore

Improved the UI of the ChatPage, Prompt and Sidebar

Implemented image detection and validation

Cleaner architecture for the project

Added PWA support

Refactor chat handling and UI improvements

Refactor ChatPage, Prompt, Sidebar, and Sitemap for improved UI and functionality

Added new landing page design

Added new chat page design

Update project structure and enhance functionality

Enhanced error handling and logging in various API routes and utility functions.

Introduced Cloudinary integration for image uploads and deletions in the messaging system.