Changelog

We have just added 3 new widgets on the Analytics Dashboard for custom.typingmind.com.

These widgets provide a quick overview of the total Active Users, Estimated Tokens, and Messages Sent within a specific timeframe.

See the announcement on X

TypingMind User Group is an enhanced version of TypingMind’s user tags at custom.typingmind.com.

User Groups allow you to organize your invited users into different categories or teams, called groups. These groups can then be used to:

✅ Manage resource access: control which AI agents, prompts, and AI models should be available to each group.
✅ Set usage limits: assign specific limits, such as model usage and the number of messages a group can send, to manage your resources and cost better.

You can also enable Admin Panel access with custom roles and permissions for each user group.

Learn more: https://docs.typingmind.com/typingmind-custom/user-management/create-and-manage-user-groups

See the announcement on X

Use Azure Whisper as speech-to-text provider on TypingMind!

Here's how: Use Azure Whisper

Eleven Flash v2 and Eleven Flash v2.5 can now be accessed on TypingMind via Text-to-Speech!

Set up Text-to-speech

Follow us on X, LinkedIn, Facebook, to stay informed about the latest updates, tips, and tutorials.

Gemini 2.0 Flash Thinking Experimental is now available on TypingMind!

Follow us on X, LinkedIn, Facebook, to stay informed about the latest updates, tips, and tutorials.

A brand new look for TypingMind sidebar and workspace bar: Sleeker, more intuitive!

Follow us on X, LinkedIn, Facebook, to stay informed about the latest updates, tips, and tutorials.

OpenAI o1 is now available on TypingMind!

• Plugins and vision support

• Faster than o1-previewUse directly via OpenAI API (usage tier 3-5) or through OpenRouter.

  

  Follow us on X, LinkedIn, Facebook, to stay informed about the latest updates, tips, and tutorials.

Now you can Chat with Videos on TypingMind.com! 🎬Directly upload your videos and analyze their content effortlessly!Note: currently, only Gemini models support chatting with videos.👉 Learn more: http://docs.typingmind.com/chat-with-files/chat-with-videos

Follow us on X, LinkedIn, Facebook, to stay informed about the latest updates, tips, and tutorials.

The new Gemini 2.0 Flash model is now available!

Follow us on X, LinkedIn, Facebook, to stay informed about the latest updates, tips, and tutorials.

Two new models:

• GPT-4o (2024-11-20): better at creative writing

• Gemini Experimental 1206: better at programming/math tasks…now available on TypingMind.com! 🔥

  

  Follow us on X, LinkedIn, Facebook, to stay informed about the latest updates, tips, and tutorials.

Cost estimation for custom models is now supported!Simply enter the price per 1M tokens for each custom model, and the app will automatically calculate the cost while you interact with it!

The latest experimental Gemini model - Gemini Experimental 1121 is now available on TypingMind!

• Significant gains on coding performance

• Stronger reasoning capabilities

• Improved visual understanding

It’s here - the biggest deal of the year! 🎉🔥 55% OFF on All Lifetime Plans!Use code: BLACKFRIDAY2024Apply to all license typesLimited time. Grab it now!

Re-arrange the pinned AI Agent on the chat interface.

• Click on the Edit button

• Drag and drop to re-order pinned AI Agents

The new Gemini Experimental 1114:

• 32k context window

• Good at understanding and aligning with your long & complex instructions.

• Ranks joint #1 overall on Chatbot Arena (even beating GPT-4o and o1-preview.)

  

  Follow us on X, LinkedIn, Facebook, to stay informed about the latest updates, tips, and tutorials.

We are excited to share with your our new Roles and Permissions system - the advanced system that allows you to add team members to your chat instance, assign specific roles and permissions, and enable them to access the Admin Panel to support you in their designated functions.

Instance owners can create Custom Roles within each chat instance. You can define a variety of roles, each with a unique set of permissions, and then assign these roles directly to team members.

TypingMind permissions determine what actions each role can perform within the Admin Panel. These permissions are assigned to roles, not individual users.

Learn more: https://docs.typingmind.com/typingmind-custom/user-management/roles-and-permissions-101

Perplexity Search Plugin on TypingMind now supports citations - which provides sources of information that the AI model generates within the response, ensuring more accurate and relevant answers.

To get it to work, you just need to enable Perplexity Search Plugin and provide relevant prompts that help trigger the plugin.

Tip: Include “Perplexity Search” and “citations” keywords in your prompt to ensure the AI model triggers the right plugin and generates responses with citations.

Now you can search the AI model you want to start a conversation with quickly.

• Click on the Model selection bar on the top of the conversation

• Type keywords to search for the model you want to chat with.

Now you can assign a specific AI Agent to your Project folder to enhance the productivity by minimizing the steps to select AI Agent within Project.

• Click to open the Project folder settings

• Select an AI agent to assign to your project. Once you select an AI Agent, new chats within that project will start with the selected AI agent. You can still change the AI agent for each individual chat if needed. Note: if the AI agent already has an assigned model, it will override the starting model setting.

Learn more Project Folders

Claude 3.5 Haiku is now available on TypingMind!Haiku is fast and particularly strong at coding. It is stated to outperform state-of-the-art models - including GPT-4o.Plugins supported. Try now!

We have just added a Share button right next to your chat activity in the left panel so you can quickly share chats.

The new Share Chat button right next to your chat activity allows you to Share chat more easily.

Easily organize your chats by:

• Recent Chats First

• Oldest Chats First

• Title A to Z

• Title Z to AJust click the sort icon in your left-side panel to choose your preference.

  

  Follow us on X, LinkedIn, Facebook, to stay informed about the latest updates, tips, and tutorials.

Edit in Canvas on TypingMind is great for editing and revising the AI response to improve AI-generated responses that meet your goals without switching back-and-forth within your conversation.Edit in Canvas can make your work smoother and more effective by allowing you to:

• **Edit text or code directly: **make changes instantly, add or adjust your text or code.

• Rewrite specific parts of your text or the entire message: make your text longer, shorter, friendlier or enter custom prompts to adjust tone, voice, style of your text.

• Use markdown formatting: you update basic styles for your text including bold, italics, headers, lists, quotes, and more.

  

  Follow us on X, LinkedIn, Facebook, to stay informed about the latest updates, tips, and tutorials.

Now you can export an AI Agent as JSON file so you or your team can easily import it to their workplace.

• Go to AI Agent

• Click the Share icon on the bottom of each AI Agent to share as JSON

  

• Click on the drop-down icon next to the Create AI Agent button and click “Import from JSON” to import the AI Agent.

  

Follow us on X, LinkedIn, Facebook, to stay informed about the latest updates, tips, and tutorials.

🔥 What’s News?

We have just added an option to hide List more button. Please note that this option only appears if the AI response includes a list of items.

🏁 How it works

1. Ask the AI assistant to give a list of anything you want.
2. The List more button appears → Click on the Square icon next to it and unpin List More option.

See the announcement on X

New plugin added to our plugin store: Firecrawl Web Page Reader.

🔥 What’s News?

Firecrawl Web Page Reader allows you to read and analyze a web page by scraping its content using the Firecrawl Scrape API.

❗ Important Notes

Rate Limits

There are rate limits for API requests depending on different pricing plans. Learn more at https://docs.firecrawl.dev/rate-limits.

Pricing Plans

Firecrawl offers various subscription types. Make sure to review the available features and limitations in your current plan. See more details at https://www.firecrawl.dev/pricing.

🏁 How it works

1. Go to Plugin Store within the Plugins menu on the left side panel
2. Click to add Firecrawl Web Page Reader and follow the instructions below:

https://docs.typingmind.com/plugins/set-up-firecrawl-web-page-reader

See the announcement on X

We have just released the Project Folders for TypingMind to help you manage and organize your work much more easily.

🔥 What’s News?

Project Folders are an improved version of the old “Folders” feature. This is built to help you better organize your projects by giving you a dedicated space to:

• Assign a specific AI model to each project.
• Add custom system instructions.
• Upload relevant documents directly to the project.

The new Project Folders allow every chat or task within a project has the right context and resources available.

🏁 How it works

1. Navigate your left sidebar and click on “+” icon to create a project folder
2. Click on Project Settings to:
   • Assign a chat model for your project
   • Provide context and custom instructions
   • Upload training files (if any)

Learn more: https://docs.typingmind.com/chat-management/project-folders

See the announcement on X

New updates for Analytics Page on custom.typingmind.com!

With this update, you can:

✅ Select custom date range to view reports.
✅ Export data to CSV within a selected period

🤔 Why This Matters

This allows you to optimize your reports for precise analysis and easily compare data across various timeframes to identify…

See the announcement on X

We have just improved UX/UI for our AI Agent page to help you better navigate the AI Agents!

✨ Changes

• Navigate AI Agent categories more easily with a sleek, modern design!
• Designed to boost your productivity and improve the usability of the AI Agent while using the app.

See the announcement on X

We have just added the new Claude 3.5 Sonnet model (claude-3-5-sonnet-20241022)!

The new model is smarter, more capable, and still at the same cost!

Plus, you can easily build powerful AI Agents on TypingMind with this upgraded model.

More details on the model: New Claude 3.5 Sonnet

See the announcement on X

Gemini Context Caching is now available on TypingMind! 🔥

Supported models:

• Gemini 1.5 Pro
• Gemini 1.5 Flash

The minimum input token count for context caching is 32,768.

🏁 How it works

• Go to Model Settings
• Expand the Advanced Model Parameter
• Scroll down to enable the “Prompt Caching” option

Learn more https://docs.typingmind.com/prompts/automatic-prompt-caching

See the announcement on X

We have just improved math syntax display on TypingMind!

Support both \(, \[, and $$ syntax when displaying math expressions.

See the announcement on X

Our Plugin Store just got more powerful!

We’ve added 3 new plugins:

🔸Slack Message Notifier (OAuth): with this plugin, you can send messages from TypingMind to a Slack channel once you authenticate with your Slack account.

🔸Slack Webhook Message Notifier: this plugin also helps send messages to a Slack channel using an incoming webhook.

🔸Google Calendar (OAuth): this allows the AI to access the user’s calendar via Google API.

See the announcement on X

Groq Whisper for Speech-to-Text (Audio Input) is now available on TypingMind!

Supported models:

🔸Whisper large-v3: provide high accuracy for multilingual transcription tasks.

🔸Distill-Whisper English: provide faster, lower cost English speech recognition while maintaining comparable accuracy.

🏁 How it works

• Click the microphone icon within the message area
• Select Speech-to-Text provider as Groq Whisper
• Enter your Groq API Key and click Check key
• Toggle the Advanced Options to select a Whisper model and enter the Groq Whisper prompt

See the announcement on X

New ElevenLabs Text-to-speech models - Eleven Turbo v2.5 and Eleven Turbo v2 are now available!

• Eleven Turbo v2.5
  • Support 32 languages (with new languages such as Vietnamese, Hungarian, and Norwegian)
  • Latency is as low as 300ms - ideal for real-time interactions (300% faster than Multilingual v2)
  • 50% cheaper than previous models
• Eleven Turbo v2
  • Similar in performance to Turbo v2.5 but focused exclusively on English, making it ideal for English-only use cases where speed is critical.

See the announcement on X

Now you can set up Custom JSON model parameters for AI Agents!

This allows you to set up custom parameters that are not available in the app.

• Custom parameters are added to the request body of the API call
• This setting may overrides some of the available parameter settings if the parameter name matches.
• Make sure that the model you are using supports the parameters you are adding

For example: to set up max tokens for AI Agent using Azure OpenAI, use: { "max_tokens": 2000 }

See the announcement on X

Upload documents to help the AI agent understand your domain knowledge better!

🚀 How it works?

• Open the AI Agent editor
• Toggle the Training files within the knowledge section
• Upload your training files for the AI Agent

Here are some notes:

• Assign a model to your AI Agent before adding training files.
• Your uploaded documents will be included directly to the model's system instruction.
• Make sure your document fits within the context length of the chosen model.

See the announcement on X

Improve UX / UI for our plugin setting page!

🔥 What’s News?

• New Plugin Store to browse plugins more easily
• Quickly install multiple plugins
• Search to find plugins faster
• Duplicate, share, edit or uninstall plugin with ease
• Improved UX/UI for a more intuitive and streamlined experience

🏁 How it works

• Go to Plugin settings on the left side panel to see the new updates.

See the announcement on X

You can now set up Custom Parameters for Specific Models!

Customizing parameters for each AI model allows you to optimize their performance based on their specific strengths and your needs. This allows you get the best possible results from each model.

🏁 How it works

• Go to Model settings on the left side panel.
• Click on the AI model you want to assign custom parameters to.
• Switch to Parameters tab
• Toggle on Override Global Model Parameters to create a custom preset for that AI model. This way, it won't use your global settings.
• Set up the parameters value as you want. (Context limit, Temperature, Top-p, Top-k, Max tokens, etc.)

Learn more: https://docs.typingmind.com/parameter-settings/custom-parameters-for-each-model

See the announcement on X

🔑You now have the option to enable or disable the Email Login for your members!

This way, your members can log in only through your preferred method, such as SSO or OAuth 2.0.

See the announcement on X

OpenAI Prompt Caching is now available on TypingMind!

Prompt Caching allows users to make repeated API calls more efficiently by reusing context from recent prompts.

• Supported on the latest version of GPT-4o, GPT-4o mini, o1-preview, o1-mini.
• You can get a 50% discount on input tokens when using cached prompts for OpenAI models. Plus, it can also reduce up to 80% in latency!

🏁 How it works

You don't need to do anything to enable OpenAI Prompt Caching - it will be automatically applied when using supported models.

Learn more on https://docs.typingmind.com/prompts/automatic-prompt-caching-(claude-and-openai)

See the announcement on X

📩 New Email Reports for custom.typingmind.com!

If you enable this option, you will receive a daily report email containing the users' messages within that day!

See the announcement on X

🟢 TypingMind Plugins now support OAuth 2.0 authentication!

This means you can now create plugins that use the user's account to authenticate instead of your fixed API key.

This authentication type makes it easy to work with external services that require OAuth.

🏁 How it works

👉To help you set up OAuth in your plugin: we've prepared a comprehensive guide: https://docs.typingmind.com/plugins/oauth-for-plugin

See the announcement on X

✅ We have launched a new Model Settings page for typingmind.com!

🔸Search and filter model - help to find the models you need quickly
🔸Easily prioritize and reorder your AI models.
🔸Get detailed model information.
🔸A cleaner layout helps you adjust settings even more easily.

🏁 How it works:

To help you clearly understand how to use our new model setting, check out here:

https://docs.typingmind.com/chat-models-settings/manage-chat-models

See the announcement on X

🧩 New Gemini Models are now available on TypingMind!

We have just updated these new Gemini models: on TypingMind:

• Gemini Flash 8B Experimental 0924
• Gemini 1.5 Flash 002
• Gemini 1.5 Pro 002

❓Why This Matters

New models bring better performance and better experience:

• Rate limits increased: 2x for Flash, 3x for Pro.
• 2x faster output and 3x lower latency
• Improved performance for tasks like math, vision, and code generation.

Easy access Google's Gemini AI models with typingmind.com:

https://docs.typingmind.com/chat-models-settings/use-with-gemini

See the announcement on X

We are thrilled to announce that you can now seamlessly integrate and deploy Anthropic AI models using AWS Bedrock directly within TypingMind! This powerful integration allows you to harness advanced AI capabilities to enhance your applications and workflows.

With this update, you can:

• Easily Connect to Anthropic Models via AWS: use the custom model option to create the AWS Anthropic models on TypingMind.
• Scale Effortlessly with AWS: Benefit from AWS's robust infrastructure to scale your AI solutions as your needs grow.
• Secure and Compliant Deployment: Utilize AWS's security features to keep your data protected and comply with industry standards.

🤔 Why This Matters

Whether you're developing chatbots, virtual assistants, or any language-driven solution, this integration accelerates your development cycle and improves performance.

🏁 How it works

To help you set up and start using this new option, we've prepared a comprehensive guide:

🔗 Use AWS Bedrock Anthropic

Set up Claude via Bedrock on TypingMind

See the announcement on X

Enable LLMs to generate citations in their responses on custom.typingmind.com!

By enabling "Show reference sources of training documents to users", the AI model will provide a link to the source within your training document.

🤔 Learn more here: https://docs.typingmind.com/typingmind-custom/branding-and-customizations/enable-llms-to-cite-sources-when-using-rag

See the announcement on X

🧩 TypingMind now allows you to create "flows" that involve multi-step, multi-agents in a single conversation to achieve a specific task.

• You can create multi-step prompt chains that tag multiple AI agents into the flow. (With each agent can have their own model, parameters, dynamic context, system prompt, plugins, etc., and have access to all previous messages in the current conversation.
• Save your flows as text in your prompt library for easy access and reuse (use slash "/" to call a prompt in the text input)
• This works in regular messages too, not just prompts, just type @ in your input and call up any AI agent in any conversation.

🏁 How it works

👉 Here is some examples on how to use the “flows”:

output two random number

Hi @[Math Professor] please calculate the sum

Hi @[Blog Image Generator] please create an image with that

Find out more about Prompt Chaining here: https://docs.typingmind.com/prompts/prompt-chaining-automated-workflow-to-enhance-model-reasoning

See the announcement on X

🧩 The new OpenAI models O1 Preview and O1 Mini are now available on typingmind.com 🔥

• New OpenAI o1-preview and o1-mini models focus on enhanced reasoning capabilities, allowing them to tackle more complex problems in science, math, and coding.
• o1-preview excels in deep reasoning tasks, while o1-mini is faster and more cost-effective, particularly for coding, being 80% cheaper.
• Both models are useful for developers and researchers needing accurate multi-step workflows or debugging, with o1-mini targeting scenarios requiring reasoning but not extensive world knowledge.
• Please note that the two new models do not support plugins

❓ Note

If you are using these models via Custom Models, make sure to untick the "Streaming Output" and "System Role" when adding the model, as these new models do not support those two features.

See the announcement on X

🧩 We’ve just improved UX/UI for the Chat Input Editor to bring you a better chat experience!

The new design is cleaner and more organized with:

• Reorganized action buttons/icons
• Added an option to adjust chat width.

➡️ Get started with TypingMind to get familiar with the interface: https://docs.typingmind.com/getting-started/get-started-with-typingmind

See the announcement on X

🧩 We have just improved the sidebar to make it easier for you to navigate your settings on typingmind.com!!

The new sidebar has been updated with new features available:

• New profile panel
• New "Workspace" bar
• Reorganized important screens.

Learn to customize your powerful sidebar: https://docs.typingmind.com/general-settings/customize-app-appearance

See the announcement on X

✅ New experimental Gemini models are now available on typingmind.com!

We have just updated 3 new experimental Gemini models on our platform:

• A new smaller variant, Gemini 1.5 Flash-8B
• A stronger Gemini 1.5 Pro Experimental 0827
• A significantly improved Gemini 1.5 Flash Experimental 0827

Easy access Google's Gemini AI models with typingmind.com: https://docs.typingmind.com/chat-models-settings/use-with-gemini

See the announcement on X

🔄Directory Sync (SCIM v2) is now available on custom.typingmind.com

You can now sync users automatically from your team's identity providers (Okta, Azure Directory, etc.) to TypingMind along with SSO.

⇒ Read our document/guide here: https://docs.typingmind.com/typingmind-custom/user-management/setup-directory-sync-(scim-v2)

See the announcement on X

✅ We have just implemented Artifacts in TypingMind - a new feature built to simplify and enhance your workflow!

TypingMind Artifacts are designed to seamlessly integrate with your existing workflow.

It is compatible with all plugins that render markdown or HTML. Also, by using Artifacts on TypingMind, you are not limited to just the Claude models but you can use with any models that support plugins.

🏁 How it works:

Artifacts in TypingMind are automatically triggered for generating structured content (e.g., code, documents, data visualizations). Therefore, you do not need to manually start an Artifact; it appears when the context requires it.

The AI will automatically deploy relevant plugins (e.g., Interactive Canvas, Render Charts, Mermaid Diagram) to generate visual elements.

To help you clearly understand about TypingMind Artifacts, learn more: https://docs.typingmind.com/plugins/typingmind-artifacts

See the announcement on X

You now can export a single chat or multiple selected chats (at once)!

🏁 How it works:

• Choose the chat (one or some) you want to export as JSON
• Directly choose export for one chat/ To export multiple chats at once, choose bulk action → export

To help you clearly understand about sharing/export the chat, check out here:

https://docs.typingmind.com/chat-management/shareexport-a-chat

See the announcement on X

The Usage Limits system for custom.typingmind.com has been revamped! 🚀

Now admins can set specific usage limits - Message and Character Limits - at different levels for better control:

🔸 Global Limits
🔸 AI Model Limits
🔸 AI Agent Limits
🔸 User Limits

See the announcement on X

Set your local Whisper endpoint for voice input on http://typingmind.com!

🏁 How it works

• Go to Settings
• Select Proxy & Org ID
• Change the OpenAI Whisper Endpoint to your local endpoint.

See the announcement on X

Automatic Prompt Caching for Claude is now available on typingmind.com!

• Prompt Caching allows Claude models to store and reuse stable contexts, such as system instructions or background information.
• With this, you can provide Claude with more extensive background knowledge and example outputs – all while cutting costs by 90% and speeding up responses by 85%!

🏁 How it works:

• Go to Model Settings
• Expand the Advanced Model Parameter
• Scroll down to enable the “Prompt Caching” option

Notes:

• The cache has a 5-minute lifetime, refreshed each time the cached content is used.
• Avoid using Prompt Caching with Dynamic Context, as changing system prompts cannot be cached.

To help you clearly understand about Prompt Caching, check out here:

https://docs.typingmind.com/prompts/automatic-prompt-caching

See the announcement on X

✅ You can now customize the left sidebar's width to better suit your workspace preferences!

Custom sidebar width: adjust the left sidebar to your preferred size.

See the announcement on X

Cost estimation for Gemini models is now available!

⚙️ How it works?

Chat with Gemini models and the cost estimation will appear on the top right corner of the app.

See the announcement on X

Added OAuth 2.0 to custom.typingmind.com 😄

Now you can integrate with ANY OAuth 2.0 provider so your users can log in more easily.

👋 Go to Admin Panel → User Management → Authentication → OAuth 2.0 to set up.

See the announcement on X

Added an option to hide the chat interface behind the login screen for custom.typingmind.com!

🔛 When this option is enabled, members will need to log in before they can access the chat interface.

See the announcement on X

✅ You can now create a workflow with multiple queued messages for AI agents on typingmind.com:

• Queue multiple messages in one go!
• This feature also works with your prompt saved in the Prompt Library.

🏁 How it works:

Use "---" to queue your messages.

See the announcement on X

✅ Gemini 1.5 Pro Experimental 0801 is now available on TypingMind.com 🔥

New Gemini models: Gemini 1.5 Pro Experimental 0801

This model currently ranked #1 on LMSYS Chatbot Arena Leaderboard!

See the announcement on X

New sidebar on custom.typingmind.com!

We have made it even easier for Admins to view and set up your chat instance with a new sidebar:

• Cleaner design
• Toggle between sections to see relevant information instantly.

See the announcement on X

No more waiting for AI replies because now you have Queue inputs on TypingMind.

Once the current AI response is finished, the AI assistant will automatically process each input one by one.

🏁 How it works:

• Submit the initial input
• While the AI assistant is answering, type more inputs and press Enter to line them up in a queue
• The AI will process each input one by one.

See the announcement on X

TypingMind is now SOC2 compliant! 🛡️

We're committed to protecting your data and privacy.

This certification validates our security practices, giving you peace of mind while using our advanced AI chat interface 🥳

Learn more: https://docs.typingmind.com/security-and-compliance

❓ What Does It Mean?

• As a customer: Not much changed, we have always been taking your data security and privacy very seriously. This certificate is just an additional proof from an external auditor to verify that we are using the best practices and industry standards regarding data security to protect your data. We hope this bring you more trust in our platform!
• As a reseller: When talking to potential clients, your clients may want to run a security audit of your vendors (in this case: TypingMind Custom). Our SOC2 certificate means that from now on, your job will be much easier. We provide our security reports and SOC2 report on request (with NDA signed). If you need one, just contact us via support@typingmind.com

See the announcement on X

✅ GPT-4o mini is now available on TypingMind.com 🥳

GPT-4o mini is now available: GPT-4o Mini is a smaller, optimized version of the GPT-4o language model designed to run efficiently on mobile devices like smartphones and tablets.

See the announcement on X

✅ You can now see the initial cost estimation before sending a prompt.

🔸You can now view the initial cost estimation before sending a prompt. This helps to estimate the cost exactly!

See the announcement on X

✅ Claude 3.5 Sonnet max output has been doubled!

Claude 3.5 Sonnet now has 8k max output. It updated from 4096 tokens to 8192 tokens.

❓ How To Use:

Quick change the token limit of Claude 3.5 Sonnet:

• Go to Advanced Model Parameters > Set the Max tokens limit to 8192.

See the announcement on X

New in custom.typingmind.com: Suspend members🛑

Members who are suspended:

• Cannot access the chat instance, but their data will be retained.
• Do not count towards the total number of seats.
• Can be unsuspended at any time.

See the announcement on X

✅ We’ve updated a new plugin on TypingMind: Mermaid diagram

Mermaid Diagram: New Plugin on typingmind.com.

• This plugin helps you generate and render diagrams using the Mermaid.js library.
• This can be used to render various types of diagrams, including Flowchart, Sequence Diagram, Class Diagram, Mind Map, Gantt Chart, etc.

To learn how to use plugin on TypingMind, refer to this: https://docs.typingmind.com/plugins/how-to-use-typingmind-plugin

See the announcement on X

⬆️ Admins now can quickly import members to your chat instance via your CSV file.

custom.typingmind.com

See the announcement on X

New Analytics Dashboard is now live on custom.typingmind.com! 📊

The new dashboard includes more charts and metrics for a comprehensive data overview.

See the announcement on X

✅ Added more AI Agent collections on typingmind.com!

Added more AI Agent collections on the library, you can find AI agents for every business area, including:

• Technical
• Design
• Marketing
• Sales
• Legal
• Finance
• Customer support
• and more!

You can also set categories for your AI Agent to better organize them for specific tasks.

See the announcement on X

✅ We’ve launched a new plugin: Interactive Canvas

New and powerful plugin - Interactive Canvas is now available on TypingMind.

This allows AI model to:

• Generate UI elements, games, charts, forms, etc.
• Interactive experiences for you users directly in TypingMind
• Can view and copy source (HTML/JS/CSS)

See the announcement on X

✅ Now you can set categories for your AI Agents to better organize them for specific tasks!

Besides the updated AI agents collection from the previous update, you can also set categories for your AI Agent to better organize them for specific tasks.

See the announcement on X

Google Gemini 1.5 Pro with 2M tokens is now available!

Google Gemini 1.5 Pro with 2M tokens is now available on TypingMind.com 🥳

Please note that this model is not yet publicly available to everyone. If you haven't had access yet, you can join the waitlist here: https://ai.google.dev/gemini-api

Easy access to Google's Gemini AI models with typingmind.com: https://docs.typingmind.com/chat-models-settings/use-with-gemini

See the announcement on X

🧩 The cost estimation calculator is now available!

Cost estimation calculator is now available! This calculator helps you estimate how much it will cost to use ChatGPT via the API key for your team.

Find out more here: https://custom.typingmind.com/tools/estimate-llm-usage-costs

See the announcement on X

New in custom.typingmind.com:

• All plugins are now run on the server by default
• You can optionally select a plugin to run on the client side if needed.

See the announcement on X

🧩 A new plugin is now available: Render Chart!

New plugin: Render Chart helps you render data into various types of charts.

→ Find out how to use plugin on TypingMind: https://docs.typingmind.com/plugins/how-to-use-typingmind-plugin

See the announcement on X

🧩 You can now sync API keys across devices!

You can now sync API keys across devices for a seamless AI chat experience with typingmind.com🥳

See the announcement on X

🧩 New "Query Training Data - Azure AI Search" plugin!

🔥 Introduce the new "Query Training Data - Azure AI Search" plugin!

This plugin allows you to connect to your own data storage and use it as the training data source for your AI agents!

➡️ See the tutorial here: https://www.youtube.com/watch?v=H_LnBxUsGJk

See the announcement on X

🔥Introduce the new "Query Training Data - Azure AI Search" plugin!

This plugin allows you to connect to your own data storage and use it as the training data source for your AI agents!

See the announcement on X

🧩 We’ve added new feature for users to host their own plugins on Github!

New features to allow easy plugin development: you can now hoa st your plugins on GitHub and the user can import it directly using your GitHub link.

Refer to this article to learn how to do: https://docs.typingmind.com/plugins/share-import-plugins

See the announcement on X

Now you can embed specific AI Agents as chat widgets using custom.typingmind.com

💬 Your members/customers can start chatting with the AI Agent immediately, without manually selecting it from a list!

See the announcement on X

Now you can set up Custom Parameters for each AI Agent on typingmind.com 🥳

Customize API parameters like Temperature, Top-p, Max tokens, etc., to control each AI Agent's outputs and optimize its performance!

➡️ Learn more about Parameter Settings: https://docs.typingmind.com/parameter-settings/parameter-settings

See the announcement on X

Now you can set up Custom Parameters for each AI Agent on typingmind.com 🥳

Customize API parameters like Temperature, Top-p, Max tokens, etc., to control each AI Agent's outputs and optimize its performance!

See the announcement on X

Now you can configure the Safety settings for Gemini models on typingmind.com!

The Gemini API provides safety settings that you can adjust during the prototyping stage to determine if your application requires more or less restrictive safety configuration. You can adjust these settings across four filter categories to restrict or allow certain types of content.

⚙️ How does it work?

In TypingMind: Go to Model Settings > Expand the Advanced Model Parameters > Adjust Safety settings

See the announcement on X

🧩 Now you can easily auto-archive and auto-delete archived chats.

You now have the option to Auto-archive old chats and Auto-delete archived chats after a specific period`!

See the announcement on X

🧩 Gemini 1.5 Flash is now available on typingmind.com!!!

We have just updated these new Gemini models on TypingMind: Gemini 1.5 Flash

Easy access Google's Gemini AI models with typingmind.com:

https://docs.typingmind.com/chat-models-settings/use-with-gemini

See the announcement on X

🧩 GPT-4o is now available on TypingMind!

GPT-4o is now available on typingmind.com! (Use via your API key)

❓Why This Matters

New models bring better performance and better experience:

✅ Smarter
✅ 2x faster
✅ Half the price
✅ Support Vision and Plugins

See the announcement on X

🧩 We’ve added "Negative Prompts" for Stable Diffusion v3 Image!

🛑 Added "Negative Prompts" for Stable Diffusion v3 Image!

Now you can specify what you don't want to see in the generated images.

See the announcement on X

🔍 Perplexity Search plugin is now available!

With Perplexity, you'll get:

• Better accuracy
• More detailed results

See the announcement on X

New in custom.typingmind.com: Transfer ownership

You can now easily transfer the ownership of your chat instance to another team member.

Note: Before you can set a member as the new owner, they must first be elevated to an Admin role.

See the announcement on X

🧩 We’ve added some TypingMind free tools!

New TypingMind free tools:

• AI Landing Page Feedback: get actionable insights to improve your landing page (hero section)
• AI model logos and icons: quickly access the LLM icons collection without wasting time to find it on the Internet!
• AI quiz

See the announcement on X

🧩 We’ve created our new Youtube channel!

We have just created a brand-new YouTube channel with detailed guidelines and tips for using the app. Subscribe now to stay updated!

Subscribe us on our new Youtube channel: https://www.youtube.com/@TypingMind

See the announcement on X

🧩 Stable Diffusion v3 Image is now available on typingmind.com!

We now have a built-in Stable Diffusion v3 Image plugin!

See the announcement on X

✨ Introducing Server Plugins on custom.typingmind.com

Server plugins are plugins that:

• Run on the server side
• Are hidden away from the users.

This allows Admins to enable and set up the plugin settings in advance, so your users just need to use them and don’t need to provide any plugin settings.

As a result, your plugin source code, settings, credentials, etc., will not be exposed to users, ensuring a seamless user experience.

See the announcement on X

We’ve made a major improvement around the plugin system for plugin developers and Typing Mind Custom users!

• Added Icon URL: you are no longer forced to use the “🧩” emoji as your plugin icon! Click “Edit”, then enter the Icon URL for your plugin, and it will be shown everywhere in the app.
• Added HTTP Action: this is a new way to implement your plugin without using JavaScript code. HTTP Action allows you to easily set up a simple HTTP request for external services. Post-processing engines (JSMEPath and Handlebars.js) are available to help you customize the output even more.
• Added Output Options: You can now render your plugin output directly to the user instead of giving the output to the AI models. The output can be rendered using Markdown or HTML. This gives you more flexibility to render interactive experiences to the users and, in some cases, saves more tokens!
• JavaScript code changes: The __CUSTOM_OUTPUT special property is now deprecated (replaced by the Output Options feature). If you have developed a plugin that uses this property in your JavaScript implementation, please update your plugins according to the new Plugins documentation (just updated!).

Read newly update of plugin here: https://docs.typingmind.com/plugins/build-a-typingmind-plugin

See the announcement on X

🧩 New GPT-4 Turbo are now available on TypingMind!

See the announcement on X

Claude models now support plugin that allows you to access the internet, generate images, etc.

See the announcement on X

✨ Added an option to view the estimated tokens

❓Why This Matters

See the announcement on X

Improved "AutoFill Content with AI" for AI Characters!

Now you can select an AI model to help generate content for your AI character!

See the announcement on X

Added an option to deactivate individual keyboard shortcuts!

Learn more: https://docs.typingmind.com/general-settings/keyboard-shortcuts

See the announcement on X

Gemini Ultra and Gemini 1.5 Pro are now available on http://typingmind.com! 🥳

The Gemini 1.5 Pro model supports 1M tokens context length AND vision 👁️ !!!

Learn more: https://docs.typingmind.com/chat-models-settings/use-with-gemini

See the announcement on X

Admins now can allow sign-ups using company email!

➕ Add your company's email domains
🟢 Anyone from your company can sign up using their company email

Available at custom.typingmind.com

See the announcement on X

Added 3 new built-in plugins:

• Stable Diffusion Image: generate images similar to DALL-E plugin. Powered by StabilityAI API.
• Web Page Reader: read content of web URLs.
• Market News: get news from stock/crypto market. Powered by Alpha Vantage API.

See the announcement on X

Plugins Server provides a proxy server for Typing Mind's Plugins. A proxy server is needed for complex use cases where a server is required for processing data that cannot be done from the client side in Typing Mind.

⚙️ How it works?

Plugins Server is open-source. See the code on GitHub: https://github.com/TypingMind/plugins-server

Learn more about Plugin Server: https://docs.typingmind.com/plugins/plugins-server

See the announcement on X

Track current context length usage to see how many tokens left in your conversation.

Navigate the top right circle within the app to view the indicator.

See the announcement on X

Use the search engine Kagi (kagi.com) as Search Suggestion Link!

⚙️ How it works?

• Go to Settings page
• Navigate the General tab
• Toggle the Auto suggest relevant keywords after each message and select Kagi
• Provide Kagi Session Link as the instruction provided in the app.

See the announcement on X

New UI for the setting panel!

See the announcement on X

TypingMind Extension is an advanced feature for developers to expand Typing Mind functionality by loading custom scripts to the app.

⚙️ How it works?

You can use this to:

• Access your Typing Mind's data and build a private Cloud Sync server (instead of using Typing Mind's default Cloud Sync)
• Add custom features to the app
• Customize the App UI

See the announcement on X

Claude 3 Haiku is the best fastest and cheapest model on earth right now!

See the announcement on X

Set up Multiple Profiles on TypingMind:

• Create different profiles, each with its own context & instructions.
• Assign API keys to each profile.

Ex: easily switch between profiles for:

• Personal projects and hobbies
• Work-related tasks

⚙️ How it works?

On the top left corner of the app, click Profile to set up your profiles.

See the announcement on X

Choose to include or exclude "Message Attachments" when syncing to save cloud space.
Message attachments: your uploaded files to the app such as images or documents.

⚙️ How it works?

Open Manage Cloud Sync → Toggle the Sync settings and select or deselect message attachments.

See the announcement on X

AI model management on http://typingmind.com is sooo much better now!

• Only shows the AI models you need for quick selection.
• Organize models in the priority order.

Learn more: https://docs.typingmind.com/chat-management/organize-chats

See the announcement on X

Now, you can control the maximum number of tokens generated!

This allows:

• Prevent potential cutoffs when interacting with AI models.
• Manage your API costs effectively.

Learn more on other parameters: https://docs.typingmind.com/parameter-settings/parameter-settings

See the announcement on X

CLAUDE 3 IS NOW AVAILABLE!

✅ 200K context token

✅ Vision Supported

✅ More powerful than GPT-4

✅ Both Opus + Sonnet are available

Learn more how to use Claude: https://docs.typingmind.com/chat-models-settings/use-with-anthropic-claude

See the announcement on X

Upload your audio files for transcription with OpenAI Whisper at http://typingmind.com!

You can upload:

• Recorded files, including meetings, interviews, webinars, lectures, etc.
• Voice memos
• and more

... to get your transcripts and interact with the AI models!

Learn more: https://docs.typingmind.com/voice-input-(speech-to-text)

See the announcement on X

• Manage your data in one place.
• Quickly delete unwanted data in just a click
• Option to Clear All Storage

See the announcement on X

New in custom.typingmind.com: Improved UI for Chat Logs!

It's now easier to view your user chat logs:

• List view / Detail View
• Filter your chat logs by time period

See the announcement on X

We just updated our sidebar with a new design: cleaner and easier to navigate!

What do you think?

See the announcement on X

New layout for pinned AI Characters.

See the announcement on X

Introducing Dynamic Context via API!

You can embed an API to your AI character's system prompt, enhancing its "smartness".

You can implement RAG with your own vector database in the AI character!

Learn more how to set this up: https://docs.typingmind.com/ai-agents/dynamic-context-via-api

See the announcement on X

Two new options to customize your AI characters in custom.typingmind.com:

See the announcement on X

🎨 Select a color theme for your sidebar!

Go to Settings > Appearance > Sidebar Color to change the color theme of your sidebar.

See the announcement on X

Brings multiple AI agents into one chat!

Simply mention @ the AI character name in the input box.

See the announcement on X

New Github repo for self-host users so they can update the app more easily.

See the announcement on X

Default or Compact style.

See the announcement on X

Gemini Pro and Gemini Pro Vision.

See the announcement on X

Gemini models are now natively supported in typingmind.com!

See the announcement on X

Get the AI to know your local time and date.

❓Why This Matters

See the announcement on X

Bring your AI Characters to life....

...by giving them unique voices that match their personalities 🔊🤖

Now you can assign your AI character to a specific Speech API / Voice that you prefer!

See the announcement on X

Use Shift + Click to quickly select multiple chats for bulk actions (delete/archive/move to folder)

See the announcement on X

Edit your custom model settings for quick update.

See the announcement on X

Add icon for your custom model by entering the icon URL while setting up the model.

See the announcement on X

OpenAI’s model updates for GPT-4 Turbo and GPT-3.5 (0125)

See the announcement on X

Big feature release: Chat Logs 💬

You can enable chat logs for your chat instance in Typing Mind Custom:

✅ Disabled by default (Private Mode)
✅ Enable only for anonymous users
✅ Enable for all users

Super helpful to review and improve the prompts from the chat history!

See the announcement on X

Typing Mind Custom now supports external authentication using JWT:

https://docs.typingmind.com/typingmind-custom/user-management/external-authentication-via-jwt

See the announcement on X

TypingMind now supports uploading XLSX files!

📁 Our supported file formats:

• Markdown
• PDF
• CSV
• XLSX
• TXT
• DOCX
• JSON
• HTML
• Image files (GPT-4 Vision only, such as .jpeg, .jpg, .png)

Drag - drop - ask, easy as 1 - 2 - 3!

Learn more: https://docs.typingmind.com/upload-training-data

See the announcement on X

The audio input pop-up is now draggable.

See the announcement on X

New dashboard UI

See the announcement on X

The new EU data center is now ready!

New customers who want to host their data in the EU can choose "Europe (Frankfurt)" when creating a new instance: typingmind.com/new-deployment

See the announcement on X

New in custom.typingmind.com

Admins now can control the available action buttons for messages on the chat UI.

See the announcement on X

Now, you can add Prompt for OpenAI Whisper to improve the quality of generated transcripts.

See the announcement on X

Archive chats in bulk.

Learn more: https://docs.typingmind.com/chat-management/organize-chats

See the announcement on X

Continue where you left off with “Reopen previous chat when app open.

See the announcement on X

OpenAI Text-to-speech is now available as speech provider on TypingMind (along with ElevenLabs, Web Speech API)

Learn on how to set up Text to speech: https://docs.typingmind.com/text-to-speech/set-up-text-to-speech

See the announcement on X