THX-02: 100,000,000 tokens in our first 4 days Four days ago we open-sourced HAL-X-AI/THX-02, a 4.5B decision model that thinks. Since then, more than 100,000,000 tokens have gone through THX-02. Developers are already using it for: - pull-request review and agent guardrails; - legal and public-sector documents; - bookkeeping, support triage and receipt reading. THX-02 reads text, JSON or images and returns a calibrated probability for every option you define. When a question needs a calculation or multi-step logic, it writes a short thought first. Otherwise it answers directly, in about 80 ms. It scores 65.06 on Decision Index 0.3. The API is free to try. No key is needed to start, and you can integrate it anywhere:
- Model (Apache-2.0): HAL-X-AI/THX-02 - Docs and free API: https://api.hal-x.ai/docs/thx-02 Thank you to everyone who built with THX-02 in its first days. We believe in intelligence for everyone, and with everyone.
THX-02 is here. An open-source Thinking Decision Model with a free API. No key, no sign-up.
We just released THX-02, our new 4.5B-parameter Thinking Decision Model developed at HAL-X AI.
Unlike traditional decision models, THX-02 introduces adaptive reasoning. It thinks when a decision requires deeper analysis or uncertainty is high, and responds directly when reasoning isn't needed.
Send text, documents, JSON, or images with typed questions and receive decisions with calibrated probabilities, typically in under 100 milliseconds.
What makes THX-02 different:
- Adaptive Thinking: Reasons only when necessary, combining fast decisions with deeper analysis. - Decision Index 0.3: Scored 65.06 with thinking and 63.83 without, outperforming Jev and several much larger decision models in our evaluation. - Multimodal: Supports text, structured data, and up to 4 images. - Multilingual: Supports 20+ languages, including Azerbaijani. - Fast inference: Around 50 ms for direct decisions and 79 ms with automatic reasoning on an RTX PRO 6000. - Free API: 60 decisions per minute per IP, no API key required. - Fully open-source: Model weights, reasoning adapter, runtime, and server under Apache 2.0.
THX-01 now has a free API. No key, no sign-up. Yesterday we open-sourced THX-01, our 322M decision model. Today we're opening the hosted API to everyone for free. Send a message or document with typed questions and get every answer in one forward pass, about 10 ms, with a calibrated probability for each option: curl https://api.hal-x.ai/v1/systemone \ -H "Content-Type: application/json" \ -d '{"state": "My card was charged twice for one order", "questions": {"team": {"type": "choice", "question": "Which team handles this?", "criteria": {"billing": "billing", "tech": "technical", "sales": "sales"}}}}' What it answers: choice, yes/no, score, exact number from the text (or null if it isn't there), verbatim excerpt, and "cite": true for supporting passages. - Free tier: 200 decisions per minute per IP - On our 2,843-ticket benchmark in four languages: 98.4% vs 97.4% for TypeSafe Jev, at about 10 ms instead of 331 ms - 18 trained languages - Weights are Apache 2.0: pip install thx01 Model: doofz/THX-01 Demo: doofz/THX-01-demo
We just open-sourced THX-01, a 322M decision model. Most AI decisions in production are classification in disguise: which team gets this ticket, is this email phishing, what's the total on this invoice. Calling a large LLM for each one is slow, expensive, and its confidence means nothing. THX-01 is built for exactly these decisions. You send it a message or document plus typed questions, and in one forward pass (~10 ms) it returns: - an answer to every question, with a calibrated probability for each option - exact numbers stated in the document, or null if the number isn't there - verbatim quotes and the passages that support each answer On our 4-language support-ticket benchmark it matches Claude Sonnet 5.5 (98.4 vs 98.5) and is about 150x faster. Because its confidence is calibrated, it handled about 75% of tickets automatically with zero errors. pip install thx01 doofz/THX-01