gaia-benchmark/results_public · Datasets at Hugging Face

model stringlengths 0 48	model_family stringlengths 0 49	system_prompt stringlengths 0 2.97k	url stringlengths 0 174	organisation stringlengths 0 22	score float64 0 0.78	score_level1 float64 0 0.91	score_level2 float64 0 0.78	score_level3 float64 0 0.62	date stringdate 2023-11-03 00:00:00 2025-04-28 00:00:00
AgentIM	GPT-4-turbo				0.375758	0.509434	0.360465	0.153846	2024-05-26
AgentIM v1.1	GPT-4-turbo				0.4	0.509434	0.406977	0.153846	2024-05-28
AutoGPT4	AutoGPT + GPT4	None	https://github.com/Significant-Gravitas/Auto-GPT.com	AutoGPT	0.048485	0.132075	0	0.038462	2023-11-03
Barcelona v0.1	Claude Sonnet 3.5, GPT-4o, o1				0.50303	0.622642	0.5	0.269231	2024-12-10
GPT3.5	GPT3	You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unles specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.	https://openai.com/blog/new-models-and-developer-products-announced-at-devday	GAIA authors	0.048485	0.075472	0.046512	0	2023-11-17
GPT4	GPT4	You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.	https://arxiv.org/abs/2303.08774	GAIA authors	0.060606	0.150943	0.023256	0	2023-11-09
GPT4 Turbo	GPT4	You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.	https://openai.com/blog/new-models-and-developer-products-announced-at-devday	GAIA authors	0.09697	0.207547	0.05814	0	2023-11-14
open Deep Research \| pass@1	o1		https://huggingface.co./blog/open-deep-research	HF 🤗 smolagents	0.551515	0.679245	0.534884	0.346154	2025-02-04
Auto-Deep-Research	claude-3-5-sonnet-20241022		https://github.com/HKUDS/Auto-Deep-Research	AutoAgent Team@HKU	0.551515	0.716981	0.534884	0.269231	2025-02-06
Hugging Face Agents + GPT-4o	GPT-4o		https://huggingface.co./docs/transformers/agents	Hugging Face 🤗	0.442424	0.584906	0.430233	0.192308	2024-06-26
HuggingFace Agents + Llama3-70B	Meta-Llama-3-70B-Instruct		https://huggingface.co./docs/transformers/en/transformers_agents	Hugging Face	0.169697	0.301887	0.116279	0.076923	2024-05-07
Langfun Agent v2.0	claude-3-5-sonnet-v2@20241022, gemini-1.5-pro-002		https://github.com/google/langfun		0.545455	0.603774	0.593023	0.269231	2024-12-02
Magentic-1	GPT-4o	Various. See linked code.	https://github.com/microsoft/autogen/tree/main/python/packages/autogen-magentic-one	MSR AI Frontiers	0.369697	0.54717	0.337209	0.115385	2024-10-14
Magentic-1 (o1)	o1 and GPT-4o (varies by agent)	Various. See linked code.	https://github.com/microsoft/autogen/tree/main/python/packages/autogen-magentic-one	MSR AI Frontiers	0.460606	0.566038	0.465116	0.230769	2024-10-15
Multi-Agent Experiment v0.1 (powered by AutoGen)	GPT-4-turbo	Multi-agent system. Various system prompts. See linked implementation.	https://aka.ms/gaia_multiagent_v01_march_1st	MSR AI Frontiers	0.393939	0.54717	0.383721	0.115385	2024-03-01
FRIDAY	GPT-4-turbo		https://os-copilot.github.io/	OS-Copilot	0.345455	0.45283	0.348837	0.115385	2024-01-24
GPT4	GPT4	You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.	https://arxiv.org/abs/2303.08774	GAIA authors	0	0	0	0	2023-11-09
RE-ODR					0.29697	0.339623	0.313953	0.153846	2025-02-11
Sibyl System v0.1	GPT-4-turbo		https://huggingface.co./spaces/gaia-benchmark/leaderboard		0.29697	0.433962	0.27907	0.076923	2024-05-20
TapeAgent v0.1	GPT-4o				0.339394	0.471698	0.348837	0.038462	2024-10-14
openresearch	timc			Timc	0	0	0	0	2025-02-10
Trase Agent	GPT-4o		https://trasesystems.com	Trase Systems	0.4	0.471698	0.406977	0.230769	2024-08-07
Trase Agent v0.2	Multi-Agent - Gemini Fine-tuned GPT-4o o1-preview		https://www.trasesystems.com/	Trase Systems	0.472727	0.584906	0.465116	0.269231	2024-10-16
Trase Agent v0.3	Claude, o1, Gemini		https://www.trasesystems.com	Trase Systems	0.70303	0.830189	0.697674	0.461538	2025-01-29
Warm-up Act	GPT-4-Turbo	You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible.	Model paper is currently under review and will be publicly shared as soon as review is completed.		0.175758	0.301887	0.151163	0	2024-02-22
h2oGPTe Agent v1.6.8	claude-3-5-sonnet		https://h2o.ai/platform/enterprise-h2ogpte/#AgenticAI	h2o.ai	0.636364	0.679245	0.674419	0.423077	2024-12-16
omne	o1-preview, gpt-4o				0.460606	0.603774	0.44186	0.230769	2024-10-20
gpt-3.5	gpt_3.5	test	test	test	0.024242	0.037736	0.023256	0	2024-07-24
qb_1	qw_2				0.133333	0.226415	0.116279	0	2025-02-24
TapeAgents & BrowserGym	claude-3.7-sonnet		https://github.com/ServiceNow/TapeAgents/tree/ui_demo/examples/gaia_agent	ServiceNow Research	0.557576	0.716981	0.534884	0.307692	2025-02-26
qbds	4o				0.236364	0.283019	0.255814	0.076923	2025-03-07
🦉 OWL - Roleplaying	GPT-4o and o3-mini	===== RULES OF USER ===== Never forget you are a user and I am a assistant. Never flip roles! You will always instruct me. We share a common interest in collaborating to successfully complete a task. I must help you to complete a difficult task. You must instruct me based on my expertise and your needs to solve the task step by step. The format of your instruction is: `Instruction: [YOUR INSTRUCTION]`, where "Instruction" describes a sub-task or question. You must give me one instruction at a time. I must write a response that appropriately solves the requested instruction. You should instruct me not ask me questions. Please note that the task may be very complicated. Do not attempt to solve the task by single step. You must instruct me to find the answer step by step. Here are some tips that will help you to give more valuable instructions about our task to me: <tips> - I have various tools to use, such as search toolkit, web browser simulation toolkit, document relevant toolkit, code execution toolkit, etc. Thus, You must think how human will solve the task step-by-step, and give me instructions just like that. For example, one may first use google search to get some initial information and the target url, then retrieve the content of the url, or do some web browser interaction to find the answer. - Although the task is complex, the answer does exist. If you can’t find the answer using the current scheme, try to re-plan and use other ways to find the answer, e.g. using other tools or methods that can achieve similar results. - Always remind me to verify my final answer about the overall task. This work can be done by using multiple tools(e.g., screenshots, webpage analysis, etc.), or something else. - If I have written code, please remind me to run the code and get the result. - Search results typically do not provide precise answers. It is not likely to find the answer directly using search toolkit only, the search query should be concise and focuses on finding sources rather than direct answers, as it always need to use other tools to further process the url, e.g. interact with the webpage, extract webpage content, etc. - If the question mentions youtube video, in most cases you have to process the content of the mentioned video. - For downloading files, you can either use the web browser simulation toolkit or write codes (for example, the github content can be downloaded via https://raw.githubusercontent.com/...). - Flexibly write codes to solve some problems, such as excel relevant tasks. </tips> Now, here is the overall task: <task>{self.task_prompt}</task>. Never forget our task! Now you must start to instruct me to solve the task step-by-step. Do not add anything else other than your instruction! Keep giving me instructions until you think the task is completed. When the task is completed, you must only reply with a single word <TASK_DONE>. Never say <TASK_DONE> unless my responses have solved your task.	https://github.com/camel-ai/owl	🐫 CAMEL-AI & HKU	0.581818	0.811321	0.546512	0.230769	2025-03-07
ant_test	ant_test	ant_test	ant_test	ant_test	0.012121	0.018868	0.011628	0	2025-03-09
				aa	0.115152	0.207547	0.093023	0	2025-03-10
noxrobot v0.1	noxrobot			NOX	0.09697	0.113208	0.104651	0.038462	2025-03-11
Ormind v0.1	Claude, Gemini		https://ormind.ai		0.551515	0.698113	0.546512	0.269231	2025-03-12
qbnlp	o1				0.442424	0.528302	0.453488	0.230769	2025-03-14
dynasaur_l1					0.10303	0.320755	0	0	2025-03-15
VM Agent	claude+deepseek			Void Main Lab	0.090909	0.075472	0.104651	0.076923	2025-03-16
				MBZUAI	0.018182	0	0.034884	0	2025-03-16
running-beta					0.315152	0.320755	0.337209	0.230769	2025-03-23
desearch	GPT-4o			zeelin (zeelin.cn)	0.569697	0.716981	0.581395	0.230769	2025-04-10
Langfun Agent v2.1	gemini-2.0-flash, claude-3-7-sonnet		https://github.com/google/langfun		0.715152	0.830189	0.686047	0.576923	2025-04-10
Anges	Gemini and Claude		anges.ai	Anges AI	0.6	0.660377	0.651163	0.307692	2025-04-11
DR-Huichi	QwQ-32B	"You are a reasoning assistant with the ability to perform web searches to help " "you answer the user's question accurately. You have special tools:\n\n" "- To perform a search: write <\|begin_search_query\|> your query here <\|end_search_query\|>.\n" "Then, the system will search and analyze relevant web pages, then provide you with helpful information in the format <\|begin_search_result\|> ...search results... <\|end_search_result\|>.\n\n" f"You can repeat the search process multiple times if necessary. The maximum number of search attempts is limited to {MAX_SEARCH_LIMIT}.\n\n" "Once you have all the information you need, continue your reasoning.\n\n" "Example:\n" "Question: \"Who got the first Nobel Prize in Physics?\"\n" "Assistant thinking steps:\n" "- I need to find out who was awarded the first Nobel Prize in Physics.\n\n" "Assistant:\n" "<\|begin_search_query\|>first Nobel Prize in Physics winner<\|end_search_query\|>\n\n" "(System returns processed information from relevant web pages)\n\n" "Assistant continues reasoning with the new information...\n\n" "Remember:\n" "- Use <\|begin_search_query\|> to request a web search and end with <\|end_search_query\|>.\n" "- When done searching, continue your reasoning.\n\n"	https://huggingface.co./Qwen/QwQ-32B	Individual	0.212121	0.358491	0.174419	0.038462	2025-04-14
DRP-val-v.1.0					0.460606	0.566038	0.488372	0.153846	2025-04-15
LRC-Huawei				LRC-Huawei	0.406061	0.528302	0.430233	0.076923	2025-04-16
Agents-On-the-fly	QwQ-32B			GenAI-LRC	0.406061	0.528302	0.430233	0.076923	2025-04-16
🦉 OWL - Workforce	claude-3.7-sonnet	You are a helpful assistant	https://hypnotic-mind-6bd.notion.site/OWL-Optimized-Workforce-Learning-for-General-Multi-Agent-Assistance-in-Real-World-Task-Automation-1d4004aeb21380158749c7f84b20643f?pvs=4	🐫 CAMEL-AI & HKU	0.690909	0.849057	0.674419	0.423077	2025-04-17
InfantAgent	claude-3.7-sonnet	You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.	https://github.com/bin123apple/InfantAgent	Evolution Intelligence	0.551515	0.584906	0.627907	0.230769	2025-04-18
InfantAgent-04-17	claude-3.7-sonnet	You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.	https://github.com/bin123apple/InfantAgent	Evolution Intelligence	0.569697	0.622642	0.627907	0.269231	2025-04-18
desearch	GPT-4o			zeelin（zeelin.cn）	0.606061	0.792453	0.604651	0.230769	2025-04-18
Ormind v0.2	Claude 3.7, GPT4.1, Gemini 2.5 pro		https://ormind.ai	Ormind	0.642424	0.754717	0.662791	0.346154	2025-04-18
AWorld	GPT-4o, DeepSeek V3	Please give me clear step-by-step instructions to complete the entire task. Always utilize search tools to gather general information for proper plans. If the task needs any special knowledge, let me know which tools I should use to help me get it done.	https://github.com/inclusionAI/AWorld/tree/main	Ant Group	0.69697	0.867925	0.697674	0.346154	2025-04-23
Aime_Agent_v0.1	claude-3.7-sonnet,gpt-4o-audio-preview	1. 你是一个资深的通用 Agent, 你需要完成用户的需求。 2. 你是最高管理者，任务由子 Agent 执行，但你也需要考虑单个子 Agent 能处理的会话轮次以及 Token上限，合理进行子 Agent 的拆分，注意：你自己不要调用工具。 3. 拿到需求后，要先进行plan并保存 plan，明确好完成任务需要的各个子 agent 并定义好各个子agent的输入输出，每次创建子agent前都要认真参考该plan，尽量把用户的原始需求也传给子 agent，要如果执行过程有新的问题，你可以调整plan 4. 当子 Agent 执行失败时，你需要进行 replan，重新创建子Agent，确保子Agent能够完成任务，对于同一个错误至少再尝试1次，确保能够完成任务。 5. 子Agent的执行结果，优先使用 SaveFile 保存在文件中			0.775758	0.90566	0.755814	0.576923	2025-04-23
xManus	claude-3.7-sonnet	Please answer the question below. You should: - Return only your answer, which should be a number, or a short phrase with as few words as possible, or a comma separated list of numbers and/or strings. - If the answer is a number, return only the number without any units unless specified otherwise. - If the answer is a string, don't include articles, and don't use abbreviations (e.g. for states). - If the answer is a comma separated list, apply the above rules to each element in the list.	https://github.com/adyuter/xManus		0.406061	0.811321	0.27907	0	2025-04-24
Agent2030	OPENAI				0.739394	0.867925	0.77907	0.346154	2025-04-24
desearch v0.2	GPT-4o, o1			zeelin（zeelin.cn）	0.648485	0.867925	0.627907	0.269231	2025-04-25
hey-0.3	DeepSeek, Doubao				0.254545	0.396226	0.232558	0.038462	2025-04-26
KKM Agent	Anthropic	You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.	deepseek.com		0.563636	0.528302	0.639535	0.384615	2025-04-27
Hero v0.1	gemini-2.5, claude-3.7				0.781818	0.867925	0.77907	0.615385	2025-04-27
my_agent		"You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]."			0	0	0	0	2025-04-27
Honey_agent	HF Transformers				0	0	0	0	2025-04-28
tobAI					0.418182	0.528302	0.418605	0.192308	2025-04-28
CoreS-Agent	claude-3.7-sonnet,gemini-2.5,gpt-4o	System prompt Please answer the question below. You should: - Return only your answer, which should be a number, or a short phrase with as few words as possible, or a comma separated list of numbers and/or strings. - If the answer is a number, return only the number without any units unless specified otherwise. - If the answer is a string, don't include articles, and don't use abbreviations (e.g. for states). - If the answer is a comma separated list, apply the above rules to each element in the list.	https://github.com/ZTE-AICloud/CoreS-Agent	ZTE-AICloud	0.678788	0.867925	0.686047	0.269231	2025-04-28

AgentIM

GPT-4-turbo

0.375758

0.509434

0.360465

0.153846

2024-05-26

AgentIM v1.1

GPT-4-turbo

0.4

0.509434

0.406977

0.153846

2024-05-28

AutoGPT4

AutoGPT + GPT4

None

https://github.com/Significant-Gravitas/Auto-GPT.com

AutoGPT

0.048485

0.132075

0

0.038462

2023-11-03

Barcelona v0.1

Claude Sonnet 3.5, GPT-4o, o1

0.50303

0.622642

0.5

0.269231

2024-12-10

GPT3.5

GPT3

You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unles specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.

https://openai.com/blog/new-models-and-developer-products-announced-at-devday

GAIA authors

0.048485

0.075472

0.046512

0

2023-11-17

GPT4

You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.

https://arxiv.org/abs/2303.08774

GAIA authors

0.060606

0.150943

0.023256

0

2023-11-09

GPT4 Turbo

GPT4

You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.

https://openai.com/blog/new-models-and-developer-products-announced-at-devday

GAIA authors

0.09697

0.207547

0.05814

0

2023-11-14

open Deep Research | pass@1

o1

https://huggingface.co./blog/open-deep-research

HF 🤗 smolagents

0.551515

0.679245

0.534884

0.346154

2025-02-04

Auto-Deep-Research

claude-3-5-sonnet-20241022

https://github.com/HKUDS/Auto-Deep-Research

AutoAgent Team@HKU

0.551515

0.716981

0.534884

0.269231

2025-02-06

Hugging Face Agents + GPT-4o

GPT-4o

https://huggingface.co./docs/transformers/agents

Hugging Face 🤗

0.442424

0.584906

0.430233

0.192308

2024-06-26

HuggingFace Agents + Llama3-70B

Meta-Llama-3-70B-Instruct

https://huggingface.co./docs/transformers/en/transformers_agents

Hugging Face

0.169697

0.301887

0.116279

0.076923

2024-05-07

Langfun Agent v2.0

claude-3-5-sonnet-v2@20241022, gemini-1.5-pro-002

https://github.com/google/langfun

0.545455

0.603774

0.593023

0.269231

2024-12-02

Magentic-1

GPT-4o

Various. See linked code.

https://github.com/microsoft/autogen/tree/main/python/packages/autogen-magentic-one

MSR AI Frontiers

0.369697

0.54717

0.337209

0.115385

2024-10-14

Magentic-1 (o1)

o1 and GPT-4o (varies by agent)

Various. See linked code.

https://github.com/microsoft/autogen/tree/main/python/packages/autogen-magentic-one

MSR AI Frontiers

0.460606

0.566038

0.465116

0.230769

2024-10-15

Multi-Agent Experiment v0.1 (powered by AutoGen)

GPT-4-turbo

Multi-agent system. Various system prompts. See linked implementation.

https://aka.ms/gaia_multiagent_v01_march_1st

MSR AI Frontiers

0.393939

0.54717

0.383721

0.115385

2024-03-01

FRIDAY

GPT-4-turbo

https://os-copilot.github.io/

OS-Copilot

0.345455

0.45283

0.348837

0.115385

2024-01-24

GPT4

You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.

https://arxiv.org/abs/2303.08774

GAIA authors

0

2023-11-09

RE-ODR

0.29697

0.339623

0.313953

0.153846

2025-02-11

Sibyl System v0.1

GPT-4-turbo

https://huggingface.co./spaces/gaia-benchmark/leaderboard

0.29697

0.433962

0.27907

0.076923

2024-05-20

TapeAgent v0.1

GPT-4o

0.339394

0.471698

0.348837

0.038462

2024-10-14

openresearch

timc

Timc

0

2025-02-10

Trase Agent

GPT-4o

https://trasesystems.com

Trase Systems

0.4

0.471698

0.406977

0.230769

2024-08-07

Trase Agent v0.2

Multi-Agent - Gemini Fine-tuned GPT-4o o1-preview

https://www.trasesystems.com/

Trase Systems

0.472727

0.584906

0.465116

0.269231

2024-10-16

Trase Agent v0.3

Claude, o1, Gemini

https://www.trasesystems.com

Trase Systems

0.70303

0.830189

0.697674

0.461538

2025-01-29

Warm-up Act

GPT-4-Turbo

You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible.

Model paper is currently under review and will be publicly shared as soon as review is completed.

0.175758

0.301887

0.151163

0

2024-02-22

h2oGPTe Agent v1.6.8

claude-3-5-sonnet

https://h2o.ai/platform/enterprise-h2ogpte/#AgenticAI

h2o.ai

0.636364

0.679245

0.674419

0.423077

2024-12-16

omne

o1-preview, gpt-4o

0.460606

0.603774

0.44186

0.230769

2024-10-20

gpt-3.5

gpt_3.5

test

0.024242

0.037736

0.023256

0

2024-07-24

qb_1

qw_2

0.133333

0.226415

0.116279

0

2025-02-24

TapeAgents & BrowserGym

claude-3.7-sonnet

https://github.com/ServiceNow/TapeAgents/tree/ui_demo/examples/gaia_agent

ServiceNow Research

0.557576

0.716981

0.534884

0.307692

2025-02-26

qbds

4o

0.236364

0.283019

0.255814

0.076923

2025-03-07

🦉 OWL - Roleplaying

GPT-4o and o3-mini

===== RULES OF USER ===== Never forget you are a user and I am a assistant. Never flip roles! You will always instruct me. We share a common interest in collaborating to successfully complete a task. I must help you to complete a difficult task. You must instruct me based on my expertise and your needs to solve the task step by step. The format of your instruction is: `Instruction: [YOUR INSTRUCTION]`, where "Instruction" describes a sub-task or question. You must give me one instruction at a time. I must write a response that appropriately solves the requested instruction. You should instruct me not ask me questions. Please note that the task may be very complicated. Do not attempt to solve the task by single step. You must instruct me to find the answer step by step. Here are some tips that will help you to give more valuable instructions about our task to me: <tips> - I have various tools to use, such as search toolkit, web browser simulation toolkit, document relevant toolkit, code execution toolkit, etc. Thus, You must think how human will solve the task step-by-step, and give me instructions just like that. For example, one may first use google search to get some initial information and the target url, then retrieve the content of the url, or do some web browser interaction to find the answer. - Although the task is complex, the answer does exist. If you can’t find the answer using the current scheme, try to re-plan and use other ways to find the answer, e.g. using other tools or methods that can achieve similar results. - Always remind me to verify my final answer about the overall task. This work can be done by using multiple tools(e.g., screenshots, webpage analysis, etc.), or something else. - If I have written code, please remind me to run the code and get the result. - Search results typically do not provide precise answers. It is not likely to find the answer directly using search toolkit only, the search query should be concise and focuses on finding sources rather than direct answers, as it always need to use other tools to further process the url, e.g. interact with the webpage, extract webpage content, etc. - If the question mentions youtube video, in most cases you have to process the content of the mentioned video. - For downloading files, you can either use the web browser simulation toolkit or write codes (for example, the github content can be downloaded via https://raw.githubusercontent.com/...). - Flexibly write codes to solve some problems, such as excel relevant tasks. </tips> Now, here is the overall task: <task>{self.task_prompt}</task>. Never forget our task! Now you must start to instruct me to solve the task step-by-step. Do not add anything else other than your instruction! Keep giving me instructions until you think the task is completed. When the task is completed, you must only reply with a single word <TASK_DONE>. Never say <TASK_DONE> unless my responses have solved your task.

https://github.com/camel-ai/owl

🐫 CAMEL-AI & HKU

0.581818

0.811321

0.546512

0.230769

2025-03-07

ant_test

0.012121

0.018868

0.011628

0

2025-03-09

aa

0.115152

0.207547

0.093023

0

2025-03-10

noxrobot v0.1

noxrobot

NOX

0.09697

0.113208

0.104651

0.038462

2025-03-11

Ormind v0.1

Claude, Gemini

https://ormind.ai

0.551515

0.698113

0.546512

0.269231

2025-03-12

qbnlp

o1

0.442424

0.528302

0.453488

0.230769

2025-03-14

dynasaur_l1

0.10303

0.320755

0

2025-03-15

VM Agent

claude+deepseek

Void Main Lab

0.090909

0.075472

0.104651

0.076923

2025-03-16

MBZUAI

0.018182

0

0.034884

0

2025-03-16

running-beta

0.315152

0.320755

0.337209

0.230769

2025-03-23

desearch

GPT-4o

zeelin (zeelin.cn)

0.569697

0.716981

0.581395

0.230769

2025-04-10

Langfun Agent v2.1

gemini-2.0-flash, claude-3-7-sonnet

https://github.com/google/langfun

0.715152

0.830189

0.686047

0.576923

2025-04-10

Anges

Gemini and Claude

anges.ai

Anges AI

0.6

0.660377

0.651163

0.307692

2025-04-11

DR-Huichi

QwQ-32B

"You are a reasoning assistant with the ability to perform web searches to help " "you answer the user's question accurately. You have special tools:\n\n" "- To perform a search: write <|begin_search_query|> your query here <|end_search_query|>.\n" "Then, the system will search and analyze relevant web pages, then provide you with helpful information in the format <|begin_search_result|> ...search results... <|end_search_result|>.\n\n" f"You can repeat the search process multiple times if necessary. The maximum number of search attempts is limited to {MAX_SEARCH_LIMIT}.\n\n" "Once you have all the information you need, continue your reasoning.\n\n" "Example:\n" "Question: \"Who got the first Nobel Prize in Physics?\"\n" "Assistant thinking steps:\n" "- I need to find out who was awarded the first Nobel Prize in Physics.\n\n" "Assistant:\n" "<|begin_search_query|>first Nobel Prize in Physics winner<|end_search_query|>\n\n" "(System returns processed information from relevant web pages)\n\n" "Assistant continues reasoning with the new information...\n\n" "Remember:\n" "- Use <|begin_search_query|> to request a web search and end with <|end_search_query|>.\n" "- When done searching, continue your reasoning.\n\n"

https://huggingface.co./Qwen/QwQ-32B

Individual

0.212121

0.358491

0.174419

0.038462

2025-04-14

DRP-val-v.1.0

0.460606

0.566038

0.488372

0.153846

2025-04-15

LRC-Huawei

0.406061

0.528302

0.430233

0.076923

2025-04-16

Agents-On-the-fly

QwQ-32B

GenAI-LRC

0.406061

0.528302

0.430233

0.076923

2025-04-16

🦉 OWL - Workforce

claude-3.7-sonnet

You are a helpful assistant

https://hypnotic-mind-6bd.notion.site/OWL-Optimized-Workforce-Learning-for-General-Multi-Agent-Assistance-in-Real-World-Task-Automation-1d4004aeb21380158749c7f84b20643f?pvs=4

🐫 CAMEL-AI & HKU

0.690909

0.849057

0.674419

0.423077

2025-04-17

InfantAgent

claude-3.7-sonnet

You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.

https://github.com/bin123apple/InfantAgent

Evolution Intelligence

0.551515

0.584906

0.627907

0.230769

2025-04-18

InfantAgent-04-17

claude-3.7-sonnet

You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.

https://github.com/bin123apple/InfantAgent

Evolution Intelligence

0.569697

0.622642

0.627907

0.269231

2025-04-18

desearch

GPT-4o

zeelin（zeelin.cn）

0.606061

0.792453

0.604651

0.230769

2025-04-18

Ormind v0.2

Claude 3.7, GPT4.1, Gemini 2.5 pro

https://ormind.ai

Ormind

0.642424

0.754717

0.662791

0.346154

2025-04-18

AWorld

GPT-4o, DeepSeek V3

Please give me clear step-by-step instructions to complete the entire task. Always utilize search tools to gather general information for proper plans. If the task needs any special knowledge, let me know which tools I should use to help me get it done.

https://github.com/inclusionAI/AWorld/tree/main

Ant Group

0.69697

0.867925

0.697674

0.346154

2025-04-23

Aime_Agent_v0.1

claude-3.7-sonnet,gpt-4o-audio-preview

1. 你是一个资深的通用 Agent, 你需要完成用户的需求。 2. 你是最高管理者，任务由子 Agent 执行，但你也需要考虑单个子 Agent 能处理的会话轮次以及 Token上限，合理进行子 Agent 的拆分，注意：你自己不要调用工具。 3. 拿到需求后，要先进行plan并保存 plan，明确好完成任务需要的各个子 agent 并定义好各个子agent的输入输出，每次创建子agent前都要认真参考该plan，尽量把用户的原始需求也传给子 agent，要如果执行过程有新的问题，你可以调整plan 4. 当子 Agent 执行失败时，你需要进行 replan，重新创建子Agent，确保子Agent能够完成任务，对于同一个错误至少再尝试1次，确保能够完成任务。 5. 子Agent的执行结果，优先使用 SaveFile 保存在文件中

0.775758

0.90566

0.755814

0.576923

2025-04-23

xManus

claude-3.7-sonnet

Please answer the question below. You should: - Return only your answer, which should be a number, or a short phrase with as few words as possible, or a comma separated list of numbers and/or strings. - If the answer is a number, return only the number without any units unless specified otherwise. - If the answer is a string, don't include articles, and don't use abbreviations (e.g. for states). - If the answer is a comma separated list, apply the above rules to each element in the list.

https://github.com/adyuter/xManus

0.406061

0.811321

0.27907

0

2025-04-24

Agent2030

OPENAI

0.739394

0.867925

0.77907

0.346154

2025-04-24

desearch v0.2

GPT-4o, o1

zeelin（zeelin.cn）

0.648485

0.867925

0.627907

0.269231

2025-04-25

hey-0.3

DeepSeek, Doubao

0.254545

0.396226

0.232558

0.038462

2025-04-26

KKM Agent

Anthropic

You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.

deepseek.com

0.563636

0.528302

0.639535

0.384615

2025-04-27

Hero v0.1

gemini-2.5, claude-3.7

0.781818

0.867925

0.77907

0.615385

2025-04-27

my_agent

"You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]."

0

2025-04-27

Honey_agent

HF Transformers

0

2025-04-28

tobAI

0.418182

0.528302

0.418605

0.192308

2025-04-28

CoreS-Agent

claude-3.7-sonnet,gemini-2.5,gpt-4o

System prompt Please answer the question below. You should: - Return only your answer, which should be a number, or a short phrase with as few words as possible, or a comma separated list of numbers and/or strings. - If the answer is a number, return only the number without any units unless specified otherwise. - If the answer is a string, don't include articles, and don't use abbreviations (e.g. for states). - If the answer is a comma separated list, apply the above rules to each element in the list.

https://github.com/ZTE-AICloud/CoreS-Agent

ZTE-AICloud

0.678788

0.867925

0.686047

0.269231

2025-04-28

Datasets:

gaia-benchmark
/

results_public

Dataset Card for "resultspublic"

Collection including gaia-benchmark/results_public

GAIA release