Qwen3-235B-A22B-Thinking-2507

MERA Created at 06.03.2026 21:33
0.795
The overall result
9
Place in the rating
In the top by tasks:
4
RCB
The result on the task is higher than human
The task is one of the main ones
9
ruWorldTree
The result on the task is higher than human
The task is one of the main ones
10
ruOpenBookQA
The result on the task is higher than human
The task is one of the main ones
6
ruHateSpeech
6
ruTiE
The result on the task is higher than human
The task is one of the main ones
10
USE
The task is one of the main ones
5
MathLogicQA
The result on the task is higher than human
The task is one of the main ones
9
ruMultiAr
The result on the task is higher than human
The task is one of the main ones
4
SimpleAr
10
MaMuRAMu
The result on the task is higher than human
The task is one of the main ones
10
ruCodeEval
The task is one of the main ones
+7
Hide
Weak tasks:
235
PARus
23
ruEthics
111
MultiQ
27
CheGeKa
45
ruDetox
87
BPS
+2
Hide

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.51 Accuracy
RCB 0.612 / 0.588 Accuracy F1 macro
USE 0.686 Grade norm
RWSD 0.842 Accuracy
PARus 0.822 Accuracy
ruTiE 0.95 Accuracy
MultiQ 0.506 / 0.357 F1 Exact match
CheGeKa 0.472 / 0.373 F1 Exact match
ruModAr 0.996 Exact match
MaMuRAMu 0.894 Accuracy
ruMultiAr 0.999 Exact match
ruCodeEval 0.719 / 0.857 / 0.884 Pass@k
MathLogicQA 0.997 Accuracy
ruWorldTree 0.994 / 0.994 Accuracy F1 macro
ruOpenBookQA 0.96 / 0.96 Accuracy F1 macro

Evaluation on open tasks:

Go to the ratings by subcategory

The table will scroll to the left

Task name Result Metric
BPS 0.983 Accuracy
ruMMLU 0.861 Accuracy
SimpleAr 1.0 Exact match
ruHumanEval 0.76 / 0.843 / 0.854 Pass@k
ruHHH 0.893
ruHateSpeech 0.928
ruDetox 0.335
ruEthics
Correct God Ethical
Virtue 0.485 0.465 0.561
Law 0.544 0.46 0.574
Moral 0.549 0.497 0.6
Justice 0.456 0.419 0.513
Utilitarianism 0.434 0.397 0.481

Information about the submission

Mera version
v1.2.0
Torch Version
2.9.1
The version of the codebase
7030e16
CUDA version
12.8
Precision of the model weights
bfloat16
Seed
1234
Batch
1
Transformers version
4.56.1
The number of GPUs and their type
8 x NVIDIA A100-SXM4-80GB
Architecture
local-chat-completions

Team:

MERA

Name of the ML model:

Qwen3-235B-A22B-Thinking-2507

Model size

235.0B

Model type:

Opened

SFT

MoE

Additional links:

https://arxiv.org/abs/2505.09388

Architecture description:

Qwen3-235B-A22B-Thinking-2507 is a large Mixture-of-Experts (MoE) decoder-only transformer language model from the Qwen3 family. The model contains 235B total parameters with about 22B active per forward pass and is optimized for complex reasoning tasks such as mathematics, science, and coding. It supports very long contexts (up to about 262K tokens) and is designed to generate intermediate reasoning steps before producing the final answer.

Description of the training:

The model follows the standard Qwen3 multi-stage training pipeline consisting of large-scale pretraining followed by post-training. Post-training includes instruction tuning and reinforcement-learning-based alignment to improve reasoning and instruction-following performance.

Pretrain data:

The model was pretrained on a large multilingual corpus of approximately 36 trillion tokens covering 119 languages. It was further post-trained on instruction-following and reasoning-oriented datasets to improve instruction following and reasoning performance.

License:

Apache License 2.0

Inference parameters

Description of the template:
{%- if tools %} {{- '<|im_start|>system ' }} {%- if messages[0].role == 'system' %} {{- messages[0].content + ' ' }} {%- endif %} {{- "# Tools You may call one or more functions to assist with the user query. You are provided with function signatures within <tools></tools> XML tags: <tools>" }} {%- for tool in tools %} {{- " " }} {{- tool | tojson }} {%- endfor %} {{- " </tools> For each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags: <tool_call> {"name": <function-name>, "arguments": <args-json-object>} </tool_call><|im_end|> " }} {%- else %} {%- if messages[0].role == 'system' %} {{- '<|im_start|>system ' + messages[0].content + '<|im_end|> ' }} {%- endif %} {%- endif %} {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %} {%- for message in messages[::-1] %} {%- set index = (messages|length - 1) - loop.index0 %} {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %} {%- set ns.multi_step_tool = false %} {%- set ns.last_query_index = index %} {%- endif %} {%- endfor %} {%- for message in messages %} {%- if message.content is string %} {%- set content = message.content %} {%- else %} {%- set content = '' %} {%- endif %} {%- if (message.role == "user") or (message.role == "system" and not loop.first) %} {{- '<|im_start|>' + message.role + ' ' + content + '<|im_end|>' + ' ' }} {%- elif message.role == "assistant" %} {%- set reasoning_content = '' %} {%- if message.reasoning_content is string %} {%- set reasoning_content = message.reasoning_content %} {%- else %} {%- if '</think>' in content %} {%- set reasoning_content = content.split('</think>')[0].rstrip(' ').split('<think>')[-1].lstrip(' ') %} {%- set content = content.split('</think>')[-1].lstrip(' ') %} {%- endif %} {%- endif %} {%- if loop.index0 > ns.last_query_index %} {%- if loop.last or (not loop.last and reasoning_content) %} {{- '<|im_start|>' + message.role + ' <think> ' + reasoning_content.strip(' ') + ' </think> ' + content.lstrip(' ') }} {%- else %} {{- '<|im_start|>' + message.role + ' ' + content }} {%- endif %} {%- else %} {{- '<|im_start|>' + message.role + ' ' + content }} {%- endif %} {%- if message.tool_calls %} {%- for tool_call in message.tool_calls %} {%- if (loop.first and content) or (not loop.first) %} {{- ' ' }} {%- endif %} {%- if tool_call.function %} {%- set tool_call = tool_call.function %} {%- endif %} {{- '<tool_call> {"name": "' }} {{- tool_call.name }} {{- '", "arguments": ' }} {%- if tool_call.arguments is string %} {{- tool_call.arguments }} {%- else %} {{- tool_call.arguments | tojson }} {%- endif %} {{- '} </tool_call>' }} {%- endfor %} {%- endif %} {{- '<|im_end|> ' }} {%- elif message.role == "tool" %} {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %} {{- '<|im_start|>user' }} {%- endif %} {{- ' <tool_response> ' }} {{- content }} {{- ' </tool_response>' }} {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %} {{- '<|im_end|> ' }} {%- endif %} {%- endif %} {%- endfor %} {%- if add_generation_prompt %} {{- '<|im_start|>assistant ' }} {%- if enable_thinking is defined and enable_thinking is false %} {{- '<think> </think> ' }} {%- endif %} {%- endif %}

System prompt:
Реши задачу по инструкции ниже. Не давай никаких объяснений и пояснений к своему ответу. Не пиши ничего лишнего. Пиши только то, что указано в инструкции. Если по инструкции нужно решить пример, то напиши только числовой ответ без хода решения и пояснений. Если по инструкции нужно вывести букву, цифру или слово, выведи только его. Если по инструкции нужно выбрать один из вариантов ответа и вывести букву или цифру, которая ему соответствует, то выведи только эту букву или цифру, не давай никаких пояснений, не добавляй знаки препинания, только 1 символ в ответе. Если по инструкции нужно дописать код функции на языке Python, пиши сразу код, соблюдая отступы так, будто ты продолжаешь функцию из инструкции, не давай пояснений, не пиши комментарии, используй только аргументы из сигнатуры функции в инструкции, не пробуй считывать данные через функцию input. Не извиняйся, не строй диалог. Выдавай только ответ и ничего больше.

Ratings by subcategory

Metric: Grade Norm
Model, team 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 8_0 8_1 8_2 8_3 8_4
Qwen3-235B-A22B-Thinking-2507
MERA
0.767 0.567 0.867 0.467 0.667 0.833 0.667 - 0.3 0.167 0.533 0.467 0.8 0.6 0.567 0.75 0.733 0.467 0.967 0.533 0.867 0.8 0.6 0.733 0.767 0.825 0.533 0.833 0.833 0.633 0.967
Model, team Honest Helpful Harmless
Qwen3-235B-A22B-Thinking-2507
MERA
0.869 0.898 0.914
Model, team Anatomy Virology Astronomy Marketing Nutrition Sociology Management Philosophy Prehistory Human aging Econometrics Formal logic Global facts Jurisprudence Miscellaneous Moral disputes Business ethics Biology (college) Physics (college) Human Sexuality Moral scenarios World religions Abstract algebra Medicine (college) Machine learning Medical genetics Professional law PR Security studies Chemistry (школьная) Computer security International law Logical fallacies Politics Clinical knowledge Conceptual_physics Math (college) Biology (high school) Physics (high school) Chemistry (high school) Geography (high school) Professional medicine Electrical engineering Elementary mathematics Psychology (high school) Statistics (high school) History (high school) Math (high school) Professional accounting Professional psychology Computer science (college) World history (high school) Macroeconomics Microeconomics Computer science (high school) European history Government and politics
Qwen3-235B-A22B-Thinking-2507
MERA
0.874 0.572 0.967 0.897 0.915 0.886 0.874 0.83 0.889 0.762 0.851 0.944 0.69 0.843 0.926 0.798 0.86 0.979 0.989 0.901 0.709 0.912 0.95 0.821 0.848 0.97 0.701 0.741 0.788 0.73 0.86 0.934 0.834 0.879 0.872 0.932 0.95 0.952 0.921 0.951 0.914 0.952 0.883 0.968 0.951 0.907 0.926 0.974 0.851 0.854 0.95 0.89 0.941 0.983 0.96 0.873 0.959
Model, team SIM FL STA
Qwen3-235B-A22B-Thinking-2507
MERA
0.751 0.624 0.75
Model, team Anatomy Virology Astronomy Marketing Nutrition Sociology Managment Philosophy Pre-History Gerontology Econometrics Formal logic Global facts Jurisprudence Miscellaneous Moral disputes Business ethics Bilology (college) Physics (college) Human sexuality Moral scenarios World religions Abstract algebra Medicine (college) Machine Learning Genetics Professional law PR Security Chemistry (college) Computer security International law Logical fallacies Politics Clinical knowledge Conceptual physics Math (college) Biology (high school) Physics (high school) Chemistry (high school) Geography (high school) Professional medicine Electrical Engineering Elementary mathematics Psychology (high school) Statistics (high school) History (high school) Math (high school) Professional Accounting Professional psychology Computer science (college) World history (high school) Macroeconomics Microeconomics Computer science (high school) Europe History Government and politics
Qwen3-235B-A22B-Thinking-2507
MERA
0.867 0.931 0.867 0.759 0.947 0.828 0.776 0.702 0.981 0.831 0.808 0.867 0.708 0.93 0.912 0.815 0.794 0.867 0.912 0.86 0.842 0.949 0.933 0.941 0.956 0.924 0.923 0.842 0.947 0.956 0.933 0.91 0.92 0.982 0.788 0.946 0.933 0.911 0.912 0.954 0.959 0.952 0.867 1 0.948 0.911 0.948 0.909 0.938 0.965 0.911 0.942 0.886 0.831 0.837 0.895 0.967
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
Qwen3-235B-A22B-Thinking-2507
MERA
0.485 0.544 0.549 0.456 0.434
Model, team Virtue Law Moral Justice Utilitarianism
Qwen3-235B-A22B-Thinking-2507
MERA
0.465 0.46 0.497 0.419 0.397
Model, team Virtue Law Moral Justice Utilitarianism
Qwen3-235B-A22B-Thinking-2507
MERA
0.561 0.574 0.6 0.513 0.481
Model, team Women Men LGBT Nationalities Migrants Other
Qwen3-235B-A22B-Thinking-2507
MERA
0.963 0.829 0.882 0.973 1 0.902