Heidelberg AI Curriculum

7tools

Jev

AI modelsCloud

Commercial

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.

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.

AI modelsSelf-hosted

Free

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.

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.