The following list contains the models we currently recommend for BYOK workspaces and the setup we . For the best user experience, we recommend to add at least the latest OpenAI GPT- and reasoning models, Claude models and Gemini models

Recommended Models

Documentation

Welcome to the product section of our knowledge base! This section is aimed at users of our platform and is designed to provide you with an in-depth understanding of each of our products.

Introduction

To select a model and interact with the messages and responses in the chat, there are a few functionalities in the chat. These are the model selector in the top left corner and the buttons to interact with prompts and responses.

Basic Chat functionalities

Canvas allows editing code or text in a dedicated screen next to a chat and iterating more easily on the output.

Canvas

If you developed a good prompt and want to reuse it in the future, there is a prompt library to store it. You can edit it, share it with others, and use it easily in a chat or with an assistant. Prompts in the prompt library are always stored within folders. You can keep your folders private or share your folders with other users in the workspace.

Prompt Library

Assistants are chatbots you can set up for a specific situation or to work with a specific document. It works similarly to the chat, with the difference that you save documents and instructions to refer to the assistant without setting the chat up again.

Introduction to Assistants

There are different tools you can ue to configure the assistant and tailor it to your specific use case.

Assistant Configuration

Langdock provides different APIs to connect models to your processes or use an assistant via the API.

API endpoints

Langdock focuses on company use cases and integrated with existing tools to fully leverage AI. This is done in two ways: By providing native integrations and by offering an easy way of building your own integrations to your custom tools.

Prompt elements are components used to craft effective prompts for interactions with LLMs. These elements help structure the prompt to ensure the LLM provides useful, relevant, and personalized responses. You don’t need to use all five elements every time, but using a few will help.

Basics: Prompt Elements

This is a basic guide to understand the fundamentals of how AI models work. It lays the foundation for deeper concepts explained in this guide.

Basics of AI models

Discover some valuable insights we have gathered over time and are excited to share with you. Dive in to discover how to craft clear, effective prompts and avoid common pitfalls, ensuring you get the best possible responses every time.

Advanced: Tips & Tricks

If you want to deepen your understanding further, we have collected some additional resources for you.

Further Resources

One of our core-values is to build a tool which is model-agnostic. That means, that we do not want to trap the user in one model, but give the choice which model to use. Each model has different strengths and we encourage you to test the different models to find the best models for your specific need.

Model Guide

Provide custom instructions to improve responses.

Custom instructions

Understand the most crucial terms in the AI world with this guide.

Dictionary

To accelerate your work with Langdock, there are a few shortcuts and tricks you can use, like CMD + K.

Tricks and Shortcuts

The Account Settings include all preferences and adjustments for a specific user. They only apply to this specific user and no one else. The account settings can be set by any user.

Account Settings

The workspace settings apply to multiple users, usually the entire workspace. This is only accessible for admins (except groups if the admins allow individual users to create groups).

Workspace Settings

Install the Langdock App in Slack and use your models and Assistants directly in Slack

Slack Bot (Setup)

Langdock has limits of prompts per user per time frame to ensure that all users can use all models at any time.

Fair Usage Policy

In the model settings, admins can add their own AI models.

Adding your own models

Rolling out AI works differently than rolling out other software tools. Here is the ideal setup we have seen in successful rollouts.

Rollout Setup

We have built the following playbook of how to roll out AI successfully based on what we have experienced with our customers. We are happy to tailor it to your individual needs and discuss your own rollout plan.

Rollout Playbook

This guide helps you to understand what great use cases are and how you identify them.

Identifying Use Cases

Here are all the steps to customize your Langdock workspace. It guides you from a blank, new workspace to be ready for onboarding your teams and for rolling out AI in your company. All of the steps are optional, but we recommend to customize the workspace to suit your technical and usability needs.

Workspace Setup

We use shared channels with our customers to have a close relation to them. This allows users to ask questions, share feedback and share use cases directly in Teams and get a fast response. We also use the channels to share product updates. This guide describes how to set up a shared Teams channel.

Shared Teams Channel

Below are all relevant links for legal and compliance teams, espcially links for security and privacy questions.

Useful Links

This section explains the different pricing of different features.

Pricing

Our invoices are created automatically in Stripe, our billing portal. In the beginning of the month, you will receive an invoice with the number of seats for the upcoming month. This calculation is corrected in the following invoice if you added or removed members in your Langdock workspace.

Invoices

Creates a model response for the given chat conversation using an OpenAI model.

OpenAI Chat completion

Creates a model response given a structured list of input messages using the Anthropic API.

Anthropic Messages

Code generation using the Codestral model from Mistral.

Codestral

Creates embeddings for text using OpenAI's embedding models

OpenAI Embeddings

The following guide explains how to create an API key in Langdock and share an assistant with the API key.

Sharing Assistants with API Keys

Creates a model response for a given Assistant.

Assistant API

Retrieve all available models for use with the Assistant API.

Models for Assistant API

Upload Attachment API

The following guide explains how to create an API key in Langdock and share a knowledge folder with the API key.

Share Knowledge Folders with the API

Upload a file to a knowledge folder

Update a file in a knowledge folder

Retrieve files from a knowledge folder

Delete a file from a knowledge folder

Search through all files in data folders shared with the API Key

Products

Resources

Settings

Admin

Platform

Support

The chat allows you to send prompts to different AI models and receive a response based on that. This guide explains how to use the model without additional tools.

Model without tools

To access specific knowledge in files, you can use the document search tool. Simply upload or connect a file and you will take the document into account.

Document Search

AI Models have a date when their knowledge ends, because they can not learn over time or save information. To access information from after that period (or simply include web results), you can use the web search tool.

Web Search

A few models are capable of processing images and taking them into account for their answer generation. This can be used to extract text, describe or analyse images.

Image Analysis (Vision)

To generate images based on your text input, you can use the image generation tool. Here, the model you selected sends a prompt to another model, which is built to generate images (most times Dall-E3).

Image Generation

To process tabular files like CSVs, Google Sheets or Excel Sheets, you can use the data analyst / python tool. Here is how to use it.

Data Analysis

Memory offers deeper personal customization of the model's behavior, by allowing them to remember information from past interactions.

Memory

Enable your users to sign in into Langdock using their Entra ID (formerly Azure Active Directory) accounts via SAML.

Microsoft Entra ID

Enable your users to sign in into Langdock using their Google accounts via SAML.

Google SAML

Provision users from Entra ID to Langdock using SCIM.

By default, users use API keys from Langdock. You can optionally use your own API keys instead.

Bring your own Keys (BYOK)

To use your own models and not the flat fee of Langdock, BYOK needs to be activated. This section guides you through the process of adding your own models.

BYOK setup

This is a short step by step guide how to set up Gemini models through Google Vertex.

Gemini Setup

1) Clear and Specific Instructions

2) Structure Your Prompt

3) Control Output Format

4) Chain Prompts

5) Context Window Tricks

6) Give the LLM a Role

Zero-shot Prompting

Few-shot Prompting

Chain-of-Thought Prompting

This guide shows you how to build an assistant. We will use an example of a job description assistant, but the steps and considerations can be used with any use case.

Creating an Assistant

We have collected a list of assistants and use cases to inspire you how to utilize Langdock for your specific needs. Let us know if you need more details or have additional requests.

Assistant Use Cases

This assistant helps you to build other assistants for specific use cases. You can use the configuration below and paste it in an assistant in your workspace.

Assistant Configurator Template

Integrations allow you to create assistants that can interact with other software tools, retrieve data, and take actions. We pre-built many new integrations for the most common software tools, but you can also build your own.

This guide shows you how to use integrations within assistants. We use the example of the Google Calendar integration, but the steps and considerations are valid for any other Langdock or custom integration.

Using Integrations

We have collected a few frequently asked questions and steps to resolve issues.

FAQ & Troubleshooting

Custom integrations allow you to integrate any tool through an API to an assistant. This gives you a huge range of opportunities. Below you find a short introduction of how to use it. We will use the example of our pre-built Google Calendar integration.

Creating Custom Integrations

This assistant can help you write actions for your integration. Simply add relevant documentation, describe your use case, and chat with the assistant to write the JavaScript.

Integration Assistant

Knowledge folders can contain up to 1,000 files and can be attached to an assistant. Below is a guide how to use this feature and where the difference is compared to attaching files directly to the assistant.

Knowledge Folders

Integrate your vector database with Langdock and let your assistant perform a semantic search across your data.

Vector Databases

Use Langdock models and Assistants directly in Slack

Model	Hosting Provider	Remarks	Hosting
GPT-4.1	Microsoft Azure	Newest GPT-4 series model	EU, US
GPT-4.1 with Canvas	Microsoft Azure	Same model as GPT-4.1, but with Canvas enabled. This separate mode is only during the Beta of the Canvas feature and will be obsolete soon, when Canvas will be recommended for more models.	EU, US
GPT-4.1 mini	Microsoft Azure	Mini version of GPT-4.1	EU (only as a global standard deployment), US
GPT-4.1 nano	Microsoft Azure	Nano version of GPT-4.1	EU, US
GPT-4o	Microsoft Azure	Normal GPT-4o model, not the latest version (08-06-2024)	EU, US
GPT-4o Mini	Microsoft Azure		EU, US
o3	Microsoft Azure	Larger reasoning model from OpenAI, available as a Global Standard model in the EU	EU (only as a global standard deployment), US
o1	Microsoft Azure	Larger reasoning model from OpenAI, available as a Global Standard model in the EU	EU (only as a global standard deployment), US
o4 mini	Microsoft Azure	Mini version of OpenAI’s o4 reasoning model	EU, US
o3 Mini	Microsoft Azure	Latest OpenAI reasoning model, standard reasoning mode - to use the standard mode, enter “o3-mini” as the model id in Langdock	EU, US
o3 Mini high	Microsoft Azure	Latest OpenAI reasoning model, high reasoning mode - to use the high reasoning mode, enter “o3-mini-high” as the model id in Langdock	EU, US
Claude Sonnet 4	Google Vertex	Latest Claude model with two modes, reasoning and no reasoning - in this version “thinking” is not activated in the Langdock interface	EU, US
Claude Sonnet 4 Reasoning	Google Vertex	Latest Claude model with two modes, reasoning and no reasoning - in this version “thinking” is not activated in the Langdock interface	EU, US
Claude 3.7 Sonnet	Google Vertex	Claude model with two modes, reasoning and no reasoning - in this version “thinking” is not activated in the Langdock interface	EU, US
Claude 3.7 Reasoning	Google Vertex	Claude model with two modes, reasoning and no reasoning - in this version “thinking” is not activated in the Langdock interface	EU, US
Mistral Large 2411	Microsoft Azure	Latest Mistral model	EU, US
Llama 3.3 70B	Microsoft Azure		EU, US
DeepSeek r1	Microsoft Azure	Larger DeepSeek reasoning model	EU, US
Gemini 2.5 Flash (Preview)	Google Vertex		US
Gemini 2.5 Flash (Preview)	Google Vertex		US
Gemini 2.0 Flash	Google Vertex		EU, US
Gemini 1.5 Pro	Google Vertex		EU, US
GPT-4.5	Microsoft Azure	Next-generation GPT model, very expensive	US
Amazon Nova Pro	AWS		US
Amazon Nova Lite	AWS		US
Flux	Black Forest Labs	Provided by either Black Forest Labs or Replicate; not offered in our Cloud at the moment	US
Sonar, Sonar Pro, Sonar Reasoning	Perplexity	Provided by Perplexity; not offered in our Cloud at the moment	US

Settings

Security

Models