Saved in:
Bibliographic Details
Main Authors: Stoianov, Dmitrii, Taranets, Danil, Tsymboi, Olga, Latypov, Ramil, Dautov, Almaz, Kruglikov, Vladislav, Surkov, Nikita, Abramov, German, Gein, Pavel, Abulkhanov, Dmitry, Gashkov, Mikhail, Zelenkovskiy, Viktor, Batalov, Artem, Medvedev, Aleksandr, Potapov, Anatolii
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.10430
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908705035911168
author Stoianov, Dmitrii
Taranets, Danil
Tsymboi, Olga
Latypov, Ramil
Dautov, Almaz
Kruglikov, Vladislav
Surkov, Nikita
Abramov, German
Gein, Pavel
Abulkhanov, Dmitry
Gashkov, Mikhail
Zelenkovskiy, Viktor
Batalov, Artem
Medvedev, Aleksandr
Potapov, Anatolii
author_facet Stoianov, Dmitrii
Taranets, Danil
Tsymboi, Olga
Latypov, Ramil
Dautov, Almaz
Kruglikov, Vladislav
Surkov, Nikita
Abramov, German
Gein, Pavel
Abulkhanov, Dmitry
Gashkov, Mikhail
Zelenkovskiy, Viktor
Batalov, Artem
Medvedev, Aleksandr
Potapov, Anatolii
contents We introduce T-pro 2.0, an open-weight Russian LLM for hybrid reasoning and efficient inference. The model supports direct answering and reasoning-trace generation, using a Cyrillic-dense tokenizer and an adapted EAGLE speculative-decoding pipeline to reduce latency. To enable reproducible and extensible research, we release the model weights, the T-Wix 500k instruction corpus, the T-Math reasoning benchmark, and the EAGLE weights on Hugging Face. These resources allow users to study Russian-language reasoning and to extend or adapt both the model and the inference pipeline. A public web demo exposes reasoning and non-reasoning modes and illustrates the speedups achieved by our inference stack across domains. T-pro 2.0 thus serves as an accessible open system for building and evaluating efficient, practical Russian LLM applications.
format Preprint
id arxiv_https___arxiv_org_abs_2512_10430
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle T-pro 2.0: An Efficient Russian Hybrid-Reasoning Model and Playground
Stoianov, Dmitrii
Taranets, Danil
Tsymboi, Olga
Latypov, Ramil
Dautov, Almaz
Kruglikov, Vladislav
Surkov, Nikita
Abramov, German
Gein, Pavel
Abulkhanov, Dmitry
Gashkov, Mikhail
Zelenkovskiy, Viktor
Batalov, Artem
Medvedev, Aleksandr
Potapov, Anatolii
Computation and Language
We introduce T-pro 2.0, an open-weight Russian LLM for hybrid reasoning and efficient inference. The model supports direct answering and reasoning-trace generation, using a Cyrillic-dense tokenizer and an adapted EAGLE speculative-decoding pipeline to reduce latency. To enable reproducible and extensible research, we release the model weights, the T-Wix 500k instruction corpus, the T-Math reasoning benchmark, and the EAGLE weights on Hugging Face. These resources allow users to study Russian-language reasoning and to extend or adapt both the model and the inference pipeline. A public web demo exposes reasoning and non-reasoning modes and illustrates the speedups achieved by our inference stack across domains. T-pro 2.0 thus serves as an accessible open system for building and evaluating efficient, practical Russian LLM applications.
title T-pro 2.0: An Efficient Russian Hybrid-Reasoning Model and Playground
topic Computation and Language
url https://arxiv.org/abs/2512.10430