Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Drex introduces four open models for probability-based decisions

AI summary

Nischay Dhankhar and eight other authors introduced Drex, a family of four open decision models that assign probabilities to user-supplied options without autoregressive decoding. The paper reports that the models were trained with supervised fine-tuning and reinforcement learning from adjusted feedback, while Drex Max scored 67.52 on Decision Index 0.4.1 and answered a typical request in 183 ms; these performance claims were not independently replicated in the supplied report.

In-depth insight

Drex offers an open alternative for classification and scoring workflows that require structured probability outputs rather than generated text, with the authors reporting lower latency and stable results under option reordering.

Computer Science > Computation and Language

arXiv:2611.07854 (cs)

[Submitted on 5 Nov 2026]

Title:Towards Foundation Decision Models: Drex

Authors:Nischay Dhankhar, Alisher Amantay, Zhanibek Datbayev, Darkhan Akhmetov, Yash Shah, Nurmukhamed Ubaidullayev, Raman Saparkhan, Dos Baha, Abulhair Saparov

View a PDF of the paper titled Towards Foundation Decision Models: Drex, by Nischay Dhankhar and 8 other authors

#Drex#Drex Max#Drex DLM#Drex v1.5#Drex v2.0#Decision Index 0.4.1#Decision Bench v4#DecideBench#Jev 1.14.0#d3-turbo#GPT-6 Sol#Claude Opus 5.5

Original · 3 min readRead the original
Group chat

No comments yet. Start the conversation.