Kimi-K2.5

MERA Created at 15.04.2026 08:11
0.746
The overall result
18
Place in the rating
In the top by tasks:
7
RWSD
The result on the task is higher than human
The task is one of the main ones
8
MultiQ
The task is one of the main ones
10
CheGeKa
The task is one of the main ones
8
ruMMLU
The result on the task is higher than human
6
ruDetox
4
ruHHH
The result on the task is higher than human
10
ruTiE
The task is one of the main ones
9
USE
The task is one of the main ones
8
MathLogicQA
The result on the task is higher than human
The task is one of the main ones
9
MaMuRAMu
The result on the task is higher than human
The task is one of the main ones
+6
Hide
Weak tasks:
153
PARus
212
ruEthics
29
ruMultiAr
46
SimpleAr
22
LCS
64
BPS
21
ruModAr
+3
Hide

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.444 Accuracy
RCB 0.603 / 0.58 Accuracy F1 macro
USE 0.689 Grade norm
RWSD 0.877 Accuracy
PARus 0.876 Accuracy
ruTiE 0.929 Accuracy
MultiQ 0.653 / 0.473 F1 Exact match
CheGeKa 0.591 / 0.478 F1 Exact match
ruModAr 0.994 Exact match
MaMuRAMu 0.9 Accuracy
ruMultiAr 0.937 Exact match
ruCodeEval 0 / 0 / 0 Pass@k
MathLogicQA 0.996 Accuracy
ruWorldTree 0.992 / 0.795 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 0.989 Accuracy
ruMMLU 0.889 Accuracy
SimpleAr 0.996 Exact match
ruHumanEval 0 / 0 / 0 Pass@k
ruHHH 0.921
ruHateSpeech 0.909
ruDetox 0.382
ruEthics
Correct God Ethical
Virtue 0.292 0.307 0.415
Law 0.307 0.293 0.421
Moral 0.326 0.32 0.445
Justice 0.262 0.271 0.381
Utilitarianism 0.259 0.294 0.356

Information about the submission

Mera version
v1.2.0
Torch Version
2.10.0
The version of the codebase
0ac3a14
CUDA version
12.8
Precision of the model weights
bfloat16
Seed
1234
Batch
1
Transformers version
4.57.6
The number of GPUs and their type
8 x NVIDIA A100-SXM4-80GB
Architecture
openai-chat-completions

Team:

MERA

Name of the ML model:

Kimi-K2.5

Model size

1,100.0B

Model type:

Opened

SFT

MoE

Additional links:

https://arxiv.org/abs/2602.02276

Architecture description:

Kimi K2.5 is an open-source, native multimodal agentic model built through continual pretraining on approximately 15 trillion mixed visual and text tokens atop Kimi-K2-Base. It seamlessly integrates vision and language understanding with advanced agentic capabilities, instant and thinking modes, as well as conversational and agentic paradigms.

License:

https://huggingface.co/moonshotai/Kimi-K2.5/blob/main/LICENSE

Inference parameters

Generation Parameters:
simplear - do_sample=false;until=[];max_gen_toks=32768;reasoning={"enabled":true,"max_tokens":10000}; \nrumultiar - do_sample=false;until=[];max_gen_toks=16384;reasoning={"enabled":true,"max_tokens":10000}; \nmultiq - do_sample=false;until=[];max_gen_toks=16384;reasoning={"enabled":true,"max_tokens":10000}; \nrumodar - do_sample=false;until=[];max_gen_toks=16384;reasoning={"enabled":true,"max_tokens":10000}; \nrucodeeval - do_sample=true;temperature=0.6;until=[];max_gen_toks=16384;reasoning={"enabled":true,"max_tokens":10000}; \nruhumaneval - do_sample=true;temperature=0.6;until=[];max_gen_toks=16384;reasoning={"enabled":true,"max_tokens":10000}; \nchegeka - do_sample=false;until=[];max_gen_toks=16384;reasoning={"enabled":true,"max_tokens":10000}; \nuse - do_sample=false;until=[];max_gen_toks=16384;reasoning={"enabled":true,"max_tokens":10000}; \nrudetox - do_sample=false;until=[];max_gen_toks=4096;reasoning={"enabled":true,"max_tokens":10000};

The size of the context:
256000

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.

Description of the template:
{%- macro render_content(msg) -%} {%- set c = msg.get('content') -%} {%- if c is string -%} {{ c }} {%- elif c is not none -%} {% for content in c -%} {% if content['type'] == 'image' or content['type'] == 'image_url' -%} <|media_begin|>image<|media_content|><|media_pad|><|media_end|> {% elif content['type'] == 'video' or content['type']== 'video_url'-%} <|kimi_k25_video_placeholder|> {% else -%} {{ content['text'] }} {%- endif -%} {%- endfor -%} {%- endif -%} {%- endmacro -%} {% macro set_roles(message) -%} {%- set role_name = message.get('name') or message['role'] -%} {%- if message['role'] == 'user' -%} <|im_user|>{{role_name}}<|im_middle|> {%- elif message['role'] == 'assistant' -%} <|im_assistant|>{{role_name}}<|im_middle|> {%- else -%} <|im_system|>{{role_name}}<|im_middle|> {%- endif -%} {%- endmacro -%} {%- macro render_toolcalls(message) -%} <|tool_calls_section_begin|> {%- for tool_call in message['tool_calls'] -%} {%- set formatted_id = tool_call['id'] -%} <|tool_call_begin|>{{ formatted_id }}<|tool_call_argument_begin|>{% if tool_call['function']['arguments'] is string %}{{ tool_call['function']['arguments'] }}{% else %}{{ tool_call['function']['arguments'] | tojson }}{% endif %}<|tool_call_end|> {%- endfor -%} <|tool_calls_section_end|> {%- endmacro -%} {# Find last non-tool-call assisitant message #} {%- set ns = namespace(last_non_tool_call_assistant_msg=-1) -%} {%- for idx in range(messages|length-1, -1, -1) -%} {%- if messages[idx]['role'] == 'assistant' and not messages[idx].get('tool_calls') -%} {%- set ns.last_non_tool_call_assistant_msg = idx -%} {%- break -%} {%- endif -%} {%- endfor -%} {# split all messages into history & suffix, reasoning_content in suffix should be reserved.#} {%- set hist_msgs = messages[:ns.last_non_tool_call_assistant_msg+1] -%} {%- set suffix_msgs = messages[ns.last_non_tool_call_assistant_msg+1:] -%} {%- if tools -%} {%- if tools_ts_str -%} <|im_system|>tool_declare<|im_middle|>{{ tools_ts_str }}<|im_end|> {%- else -%} <|im_system|>tool_declare<|im_middle|>{{ tools | tojson(separators=(',', ':')) }}<|im_end|> {%- endif -%} {%- endif -%} {%- for message in hist_msgs -%} {{set_roles(message)}} {%- if message['role'] == 'assistant' -%} <think></think>{{render_content(message)}} {%- if message.get('tool_calls') -%} {{render_toolcalls(message)}} {%- endif -%} {%- elif message['role'] == 'tool' -%} {%- set tool_call_id = message.tool_call_id -%} ## Return of {{ tool_call_id }} {{render_content(message)}} {%- elif message['content'] is not none -%} {{render_content(message)}} {%- endif -%} <|im_end|> {%- endfor -%} {%- for message in suffix_msgs -%} {{set_roles(message)}} {%- if message['role'] == 'assistant' -%} {%- if thinking is defined and thinking is false -%} <think></think>{{render_content(message)}} {%- else -%} {%- set rc = message.get('reasoning_content', '') -%} <think>{{rc}}</think>{{render_content(message)}} {%- endif -%} {%- if message.get('tool_calls') -%} {{render_toolcalls(message)}} {%- endif -%} {%- elif message['role'] == 'tool' -%} {%- set tool_call_id = message.tool_call_id -%} ## Return of {{ tool_call_id }} {{render_content(message)}} {%- elif message['content'] is not none -%} {{render_content(message)}} {%- endif -%} <|im_end|> {%- endfor -%} {%- if add_generation_prompt -%} <|im_assistant|>assistant<|im_middle|> {%- if thinking is defined and thinking is false -%} <think></think> {%- else -%} <think> {%- endif -%} {%- endif -%}

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
Kimi-K2.5
MERA
0.567 0.667 0.8 0.333 0.7 0.933 0.933 - 0.133 0.433 0.667 0.633 0.8 0.833 0.3 0.683 0.633 0.367 0.8 0.5 0.867 0.8 0.633 0.767 0.7 0.842 0.667 0.833 0.833 0.733 0.833
Model, team Honest Helpful Harmless
Kimi-K2.5
MERA
0.934 0.864 0.966
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
Kimi-K2.5
MERA
0.874 0.566 0.98 0.932 0.918 0.925 0.903 0.9 0.92 0.821 0.833 0.937 0.76 0.88 0.957 0.853 0.85 0.979 0.989 0.901 0.743 0.901 0.96 0.844 0.911 1 0.774 0.75 0.841 0.75 0.92 0.959 0.914 0.919 0.909 0.962 0.97 0.945 0.94 0.951 0.904 0.941 0.876 0.966 0.953 0.926 0.951 0.989 0.911 0.902 0.98 0.928 0.946 0.983 0.95 0.873 0.953
Model, team SIM FL STA
Kimi-K2.5
MERA
0.798 0.638 0.766
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
Kimi-K2.5
MERA
0.889 0.931 0.867 0.815 0.961 0.897 0.81 0.842 0.962 0.815 0.833 0.883 0.708 0.922 0.924 0.852 0.822 0.8 0.895 0.86 0.895 0.983 0.956 0.935 0.956 0.939 0.872 0.825 0.947 0.911 0.911 0.923 0.946 0.93 0.818 0.929 0.956 0.933 0.912 0.954 0.963 0.937 0.867 0.978 0.897 0.911 0.948 0.932 0.938 0.947 0.956 0.971 0.861 0.818 0.837 0.918 0.933
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
Kimi-K2.5
MERA
0.292 0.307 0.326 0.262 0.259
Model, team Virtue Law Moral Justice Utilitarianism
Kimi-K2.5
MERA
0.307 0.293 0.32 0.271 0.294
Model, team Virtue Law Moral Justice Utilitarianism
Kimi-K2.5
MERA
0.415 0.421 0.445 0.381 0.356
Model, team Women Men LGBT Nationalities Migrants Other
Kimi-K2.5
MERA
0.981 0.743 0.882 0.919 1 0.869