Heidelberg AI Curriculum

18tools

Locally Uncensored

View profile

Local & self-hosted appsSelf-hosted · Cloud

FreeOpen source

What you can do

Locally Uncensored is a free, open-source desktop app for Windows and Linux that runs AI models on your own computer. One program holds a chat window with document questions, image and video generation through ComfyUI, a coding agent and phone access to your own PC. The chat engine is built in, and a first-start wizard downloads a model. The maker also sells hosted graphics cards as LU Cloud.

Before choosing

No Mac desktop build. The Windows installer is not code-signed. The starter models offered at first start have their refusals removed; standard models are a separate list. Without a graphics card, answers are slow.

Jev

View profile

AI modelsCloud

Commercial

What you can do

Jev is TypeSafe AI's hosted System One model for making typed decisions inside software. Give it text or structured text state and narrow Choice, Score or Noul questions; it returns probabilities for routing, classification and scoring. It does not generate prose or code and cannot replace a chat or coding-agent LLM. The documented Jev 1.13 version is available through TypeSafe's API in early access. Start with a decision in your own workflow and keep the final action in your application code.

Before choosing

Jev is a decision model, not a generative LLM: no chat, prose generation, code writing, images, audio or video input. The publisher documents 64k tokens per request and a 32k limit for state plus the longest question for jev-1.13.0. Confidence describes the distribution and does not guarantee an individual answer is correct; the publisher documents failures on counting, date arithmetic, indirection and adversarial text. Requires TypeSafe cloud access and an API key; prices and rate limits may change. No local weights or hands-on performance evaluation were established.

Microsoft 365 Copilot

View profile

Web servicesLocal & self-hosted appsCloud · Desktop · Mobile

Free tier

What you can do

Microsoft 365 Copilot (recently renamed Microsoft Copilot) is Microsoft's AI assistant for Word, Excel, Outlook, PowerPoint and Teams. Copilot Chat, the web-grounded chat, comes free with eligible Microsoft 365 work or school plans. The paid Microsoft 365 Copilot add-on licence adds in-app drafting and automatic grounding in the organisation's own emails, files and meetings through Microsoft Graph.

Before choosing

Only Copilot Chat is free, and even then only for users signed in with a Microsoft Entra work or school account; a personal Microsoft account does not receive Microsoft's enterprise data protection. Automatic grounding in an organisation's own files, emails and meetings needs the paid Microsoft 365 Copilot add-on licence, not Copilot Chat.

DeepL

View profile

Web servicesLocal & self-hosted appsCloud · Desktop · Mobile

Free tier

What you can do

DeepL is a machine translation and writing-assistant service with a web translator, the DeepL Write editing tool, desktop apps, browser extensions and mobile apps. The free plan translates text and one document a month in the browser; paid Pro plans raise the limits, add glossaries, and guarantee submitted text is deleted after translation rather than used for training.

Before choosing

Free-plan text is used to train DeepL's models; only paid Pro plans guarantee no training and deletion of submitted text after translation.

Mammouth AI

View profile

Web servicesCloud

Commercial

What you can do

Compare responses from multiple AI models in one paid cloud app using one-click reprompting. Draft with uploaded documents, organise chats in projects, search the web, and generate images or videos within plan-dependent usage limits.

Before choosing

Official monthly prices checked 20 September 2026: Starter €12, Standard €24, Expert €72, including VAT; the homepage separately advertises from €10/month. Usage is metered, with model fallback and file limits. Mammouth claims no model training, but stores conversation history; its detailed policy permits up to 30 rolling days after deletion and describes possible transfers outside the EEA. These are vendor statements, not independently audited guarantees. No subscription or generation was tested.

Ornith-1.5-35B-A3B

View profile

AI modelsSelf-hosted

Free

What you can do

Ornith-1.5-35B-A3B is an open-weight reasoning model for coding, repository work and agents that call tools. Its Qwen3.5-family mixture-of-experts architecture contains about 35 billion parameters but activates roughly 3 billion per token. Ornith describes training that jointly improves task generation, agent scaffolds and solutions through reinforcement learning. The published context window is 262,144 tokens. It is a candidate for developers who want to run a coding assistant on their own infrastructure: the official card documents vLLM and SGLang servers with OpenAI-compatible APIs, plus GGUF use through Ollama or llama.cpp. Start with a small repository task and check the result against your tests before expanding its role.

Before choosing

About 3B active parameters describes computation per token, not the memory needed to load the model. The publisher estimates roughly 70 GB for BF16 weights; its 256K serving example uses two 80 GB GPUs. Quantization reduces weight memory, while long contexts need additional memory.