Skip to content

Commit f55fc1e

Browse files
BinWang28BinWang28
authored
Merge pull request #30 from MiroMindAI/binwang_dev
feat(mirothinker): cleanup deprecated configs and update documentation to reproduce mirothinker results
2 parents 70d934f + 3d1d194 commit f55fc1e

File tree

132 files changed

+3419
-4893
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

132 files changed

+3419
-4893
lines changed

‎.env.template‎

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -17,3 +17,7 @@ SUMMARY_LLM_API_KEY=xxxx
1717
SUMMARY_LLM_MODEL_NAME=xxxx
1818

1919
HF_TOKEN=xxxx
20+
21+
# TencentCloud credentials for Sogou search (used by serper_sogou_search tool)
22+
TENCENTCLOUD_SECRET_ID=xxxx
23+
TENCENTCLOUD_SECRET_KEY=xxxx

‎README.md‎

Lines changed: 38 additions & 153 deletions
Original file line numberDiff line numberDiff line change
@@ -1,207 +1,93 @@
11
<div align="center">
22
<img src="docs/mkdocs/docs/assets/miroflow_logo.png" width="45%" alt="MiroFlow" />
3-
</div>
4-
5-
<br>
63

7-
8-
<div align="center">
4+
<h3>Open-Source Research Agent Framework with State-of-the-Art Performance</h3>
95

106
[![DEMO](https://img.shields.io/badge/Demo-FFB300?style=for-the-badge&logo=airplayvideo&logoColor=white)](https://dr.miromind.ai/)
11-
[![MODELS](https://img.shields.io/badge/Models-5EDDD2?style=for-the-badge&logo=huggingface&logoColor=ffffff&labelColor)](https://huggingface.co/collections/miromind-ai/mirothinker-v02-68af084a18035f57b17cd902)
12-
[![DATA](https://img.shields.io/badge/Data-0040A1?style=for-the-badge&logo=huggingface&logoColor=ffffff&labelColor)](https://huggingface.co/datasets/miromind-ai/MiroVerse-v0.1)
13-
[![BLOG](https://img.shields.io/badge/Blog-4285F4?style=for-the-badge&logo=google-chrome&logoColor=white)](https://miromind.ai/blog/miroflow)
14-
15-
[![GITHUB](https://img.shields.io/badge/Github-24292F?style=for-the-badge&logo=github&logoColor=white)](https://github.com/MiroMindAI)
7+
[![MODELS](https://img.shields.io/badge/Models-5EDDD2?style=for-the-badge&logo=huggingface&logoColor=ffffff&labelColor)](https://huggingface.co/miromind-ai)
8+
[![DOCS](https://img.shields.io/badge/Docs-8CA1AF?style=for-the-badge&logo=readthedocs&logoColor=white)](https://miromindai.github.io/miroflow/)
9+
[![WEBSITE](https://img.shields.io/badge/Website-4285F4?style=for-the-badge&logo=google-chrome&logoColor=white)](https://miromind.ai)
1610
[![DISCORD](https://img.shields.io/badge/Discord-5865F2?style=for-the-badge&logo=discord&logoColor=white)](https://discord.com/invite/GPqEnkzQZd)
17-
[![RedNote](https://img.shields.io/badge/RedNote-FF2442?style=for-the-badge&logo=revoltdotchat&logoColor=white)](https://www.xiaohongshu.com/user/profile/5e353bd80000000001000239)
11+
[![RedNote](https://img.shields.io/badge/RedNote-FF2442?style=for-the-badge&logo=revoltdotchat&logoColor=white)](https://www.xiaohongshu.com/user/profile/663098830000000003033edc)
1812

1913
</div>
2014

2115
<div align="center">
16+
<strong>MiroFlow</strong> is an open-source research agent framework that achieves <strong>#1 ranking</strong> across representative benchmarks (FutureX, GAIA, HLE, xBench-DeepSearch, BrowseComp).<br>
17+
It powers <a href="https://github.com/MiroMindAI/mirothinker">MiroThinker</a>, our open-source agent foundation model with native tool-assisted reasoning.
18+
</div>
2219

23-
## 📚 **[READ THE DOCUMENTATION](https://miromindai.github.io/MiroFlow/)**
24-
25-
### 🚀 [Try Demo](https://dr.miromind.ai/)
20+
<br>
2621

22+
<div align="center">
23+
<img src="docs/mkdocs/docs/assets/futurex_results.jpg" width="100%" alt="FutureX Benchmark Results" />
2724
</div>
2825

2926
---
3027

31-
**MiroFlow** is an open-source research agent framework that achieves state-of-the-art performance on representative benchmarks (FutureX, GAIA, HLE, xBench-DeepSearch, BrowserComp). It powers [MiroThinker](https://github.com/MiroMindAI/mirothinker), our open-source agent foundation model with native tool-assisted reasoning.
32-
3328
## 📰 News
3429

35-
- **[2026-03]**: 🎉 **MiroFlow 1.6 + MiroThinker 1.6**: Major release with Web Application interface (FastAPI + React), comprehensive verifier system for benchmark evaluation, and expanded LLM support including Kimi K2.5 and GPT-5.
30+
- **[2026-03]**: **MiroFlow 1.6 + MiroThinker 1.6**: Major release with Web Application interface (FastAPI + React), comprehensive verifier system for benchmark evaluation, and expanded LLM support including Kimi K2.5 and GPT-5.
3631

3732
<details>
3833
<summary><strong>Previous Updates</strong></summary>
3934

40-
- **[2025-09-15]**: 🎉🎉 **MiroFlow v0.3**: Enhanced codebase architecture and significantly improved benchmark performance, boosting GPT-5's prediction accuracy for future events by 11%. MiroFlow now ranks #1 in the future prediction benchmark. See [FutureX](https://futurex-ai.github.io/).
41-
- **[2025-08-27]**: **MiroFlow v0.2**: Achieves state-of-the-art performance across [multiple agentic benchmarks](https://miromind.ai/blog/miroflow), including HLE (27.2%), HLE-Text-Only (29.5%), BrowserComp-EN (33.2%), BrowserComp-ZH (47.1%), and xBench-DeepSearch (72.0%).
35+
- **[2025-09-15]**: **MiroFlow v0.3**: Enhanced codebase architecture and significantly improved benchmark performance, boosting GPT-5's prediction accuracy for future events by 11%. MiroFlow now ranks #1 in the future prediction benchmark. See [FutureX](https://futurex-ai.github.io/).
36+
- **[2025-08-27]**: **MiroFlow v0.2**: Achieves state-of-the-art performance across [multiple agentic benchmarks](https://miromind.ai), including HLE (27.2%), HLE-Text-Only (29.5%), BrowserComp-EN (33.2%), BrowserComp-ZH (47.1%), and xBench-DeepSearch (72.0%).
4237
- **[2025-08-26]**: Released [GAIA Validation Trace](docs/public_trace.md) (73.94% pass@1) and [Gradio Demo](https://github.com/MiroMindAI/MiroThinker/tree/main/apps/gradio-demo) for local deployment.
4338
- **[2025-08-08]**: **MiroFlow v0.1**: Complete open-source release of the research agent framework.
4439

4540
</details>
4641

4742
---
4843

49-
## 📋 Table of Contents
44+
## Highlights
5045

51-
- 🚀 [Get Started](#-get-started-in-under-5-minutes)
52-
- 🏗️ [Architecture](#️-architecture)
53-
- 🌟 [Highlights](#-highlights)
54-
- 🔧 [Models & Tools](#-supported-models--tools)
55-
- 📈 [Benchmarks](#-performance-on-benchmarks)
56-
- ❓ [FAQ](#-faq)
57-
- 📖 [References](#-references)
46+
- **Reproducible State-of-the-Art Performance**: #1 ranking across [multiple representative agentic benchmarks](https://miromindai.github.io/miroflow/evaluation_overview/), including FutureX, GAIA, HLE, xBench-DeepSearch, and BrowseComp.
47+
- **High Concurrency & Reliability**: Robust concurrency management and fault-tolerant design for handling rate-limited APIs and unstable networks.
48+
- **Cost-Effective Deployment**: Run a research agent service on a single RTX 4090 with the open-source [MiroThinker](https://github.com/MiroMindAI/mirothinker) model and free tools.
5849

5950
---
6051

61-
## 🚀 Get Started in Under 5 Minutes
52+
## Performance on Benchmarks
6253

63-
### 📋 Prerequisites
54+
<div align="center">
55+
<img width="100%" alt="MiroThinker Performance" src="docs/mkdocs/docs/assets/mirothinker.png" />
56+
</div>
6457

65-
- **Python**: 3.12 or higher
66-
- **Package Manager**: [`uv`](https://docs.astral.sh/uv/)
67-
- **Operating System**: Linux, macOS
58+
<div align="center">
59+
<img width="100%" alt="BrowseComp MiroThinker Performance" src="docs/mkdocs/docs/assets/bc-mirothinker.png" />
60+
</div>
6861

69-
### ⚡ Quick Setup
62+
Follow our detailed guides to reproduce benchmark results in our [Benchmarks Documentation](https://miromindai.github.io/miroflow/evaluation_overview/).
7063

71-
MiroFlow provides a web interface powered by FastAPI and React for interactive agent tasks.
64+
---
65+
66+
## Quick Start
7267

7368
```bash
7469
# 1. Clone and setup
75-
git clone https://github.com/MiroMindAI/MiroFlow && cd MiroFlow
70+
git clone https://github.com/MiroMindAI/miroflow && cd miroflow
7671
uv sync
7772

7873
# 2. Configure API keys
7974
cp .env.template .env
8075
# Edit .env and add your API keys (see .env.template for details)
8176

82-
# 3. Launch the web application
83-
bash scripts/start_web.sh
84-
```
85-
86-
Access the web interface at `http://localhost:8000` and start querying the agent. The default agent configuration is `config/agent_web_demo.yaml`.
87-
88-
> **💡 Tip:** If you encounter issues, check that your API keys are correctly set in the `.env` file and that all dependencies are installed.
89-
90-
### 🖥️ CLI Usage
91-
92-
You can also run single tasks from the command line:
93-
94-
```bash
95-
# Run a single task with a custom question
77+
# 3. Run your first task
9678
bash scripts/test_single_task.sh \
9779
--config config/agent_quickstart.yaml \
9880
--task-question "What is the first country listed in the XLSX file that have names starting with Co?" \
9981
--file-path data/FSI-2023-DOWNLOAD.xlsx
10082
```
10183

102-
🎉 Expected Output: Your agent should return \boxed{Congo Democratic Republic} 😊
103-
104-
💡 Tip: If you encounter issues, check that your API key is correctly set in the .env file and that all dependencies are installed.
105-
106-
107-
---
108-
109-
## 🏗️ Architecture
110-
111-
<div align="center">
112-
<img src="docs/mkdocs/docs/assets/miroflow_architecture.png" width="100%" alt="MiroFlow Architecture">
113-
</div>
114-
115-
<details>
116-
<summary><strong>📹 Demo: Research Assistant</strong></summary>
117-
<br>
118-
<table align="center" style="border: 1px solid #ccc; border-radius: 8px; padding: 12px; background-color: #f9f9f9; width: 60%;">
119-
<tr>
120-
<td style="text-align: center; padding: 10px;">
121-
<span style="font-size: 0.9em; color: #555;">Read CVPR 2025 Best Paper and Provide Research Advice</span>
122-
<br>
123-
<video src="https://github.com/user-attachments/assets/99ed3172-6e9a-467a-9ccb-be45957fe2e4"
124-
controls muted preload="metadata"
125-
width="50%" height="50%"
126-
</video>
127-
</td>
128-
</tr>
129-
</table>
130-
</details>
131-
132-
---
133-
134-
## 🌟 Highlights
135-
136-
- **Reproducible State-of-the-Art Performance**: #1 ranking across [multiple representative agentic benchmarks](https://miromindai.github.io/MiroFlow/evaluation_overview/), including FutureX, GAIA, HLE, xBench-DeepSearch, and BrowserComp benchmarks)
137-
- **High Concurrency & Reliability**: Built with robust concurrency management and fault-tolerant design, MiroFlow efficiently handles rate-limited APIs and unstable networks, ensuring seamless trajectory collection and reliable execution of complex tasks.
138-
- **Cost-Effective Deployment**: Powered by the open-source MiroThinker model, MiroFlow can run a research agent service on a single RTX 4090. The entire stack relies on free, open-source tools, making it simple to deploy, scale, and reproduce. See [MiroThinker](https://github.com/MiroMindAI/mirothinker).
139-
140-
---
141-
142-
## 🔧 Supported Models & Tools
143-
144-
- **Models**: MiroThinker 1.6, GPT-4o, GPT-5, Claude, Gemini, Qwen, Kimi K2.5, etc.
145-
- **Tools**: [Audio Transcription](https://github.com/MiroMindAI/MiroFlow/blob/main/src/tool/mcp_servers/audio_mcp_server.py), [Python](https://github.com/MiroMindAI/MiroFlow/blob/main/src/tool/mcp_servers/python_mcp_server.py), [File Reading](https://github.com/MiroMindAI/MiroFlow/blob/main/src/tool/mcp_servers/reading_mcp_server.py), [Reasoning](https://github.com/MiroMindAI/MiroFlow/blob/main/src/tool/mcp_servers/reasoning_mcp_server.py), [Google Search](https://github.com/MiroMindAI/MiroFlow/blob/main/src/tool/mcp_servers/searching_mcp_server.py), [VQA](https://github.com/MiroMindAI/MiroFlow/blob/main/src/tool/mcp_servers/vision_mcp_server.py), [Web Scraping](https://github.com/MiroMindAI/MiroFlow/blob/main/src/tool/mcp_servers/jina_scrape_llm_summary_mcp_server.py), E2B, etc.
84+
Expected output: `\boxed{Congo Democratic Republic}`
14685

86+
See [full documentation](https://miromindai.github.io/miroflow/quickstart/) for web app setup, more examples, and configuration options.
14787

14888
---
14989

150-
## 📈 Performance on Benchmarks
151-
152-
We achieved the #1 ranking on the FutureX Benchmark Leaderboard as of September 10, 2025, boosting GPT-5's prediction accuracy for future events by 11%.
153-
154-
<div align="center">
155-
<img width="100%" alt="image" src="docs/mkdocs/docs/assets/futurex-09-12.png" />
156-
</div>
157-
158-
We benchmark MiroFlow on a series of benchmarks, including **GAIA**, **HLE**, **BrowseComp**, and **xBench-DeepSearch**, and achieved SOTA results.
159-
160-
<img width="100%" alt="image" src="docs/mkdocs/docs/assets/benchmark_results.png" />
161-
162-
| Model/Framework | GAIA Val | HLE | HLE-Text | BrowserComp-EN | BrowserComp-ZH | xBench-DeepSearch |
163-
|----------------|----------|-----|----------|----------------|----------------|-------------------|
164-
| **MiroFlow** | **82.4%** | **27.2%** | 29.5% | 33.2% | **47.1%** | **72.0%** |
165-
| OpenAI Deep Research | 67.4% | 26.6% | - | **51.5%** | 42.9% | - |
166-
| Gemini Deep Research | - | 26.9% | - | - | - | 50+% |
167-
| Kimi Researcher | - | - | 26.9% | - | - | 69.0% |
168-
| WebSailor-72B | 55.4% | - | - | - | 30.1% | 55.0% |
169-
| Manus | 73.3% | - | - | - | - | - |
170-
| DeepSeek v3.1 | - | - | **29.8%** | - | - | 71.2% |
171-
172-
Follow our detailed guides to reproduce benchmark results in our [Benchmarks Documentation](https://miromindai.github.io/MiroFlow/evaluation_overview/)
173-
174-
---
175-
176-
## ❓ FAQ
177-
178-
<details>
179-
<summary><strong>What API keys do I need?</strong></summary>
180-
<br>
181-
The required API keys depend on your configuration. See <code>.env.template</code> for the full list. Common keys include: <code>OPENROUTER_API_KEY</code> (for models via OpenRouter), <code>SERPER_API_KEY</code> (web search), <code>JINA_API_KEY</code> (web scraping), and <code>E2B_API_KEY</code> (Python code execution).
182-
</details>
183-
184-
<details>
185-
<summary><strong>Can I use other language models besides OpenRouter?</strong></summary>
186-
<br>
187-
Yes, MiroFlow supports various language models. Check our documentation for configuration details.
188-
</details>
189-
190-
<details>
191-
<summary><strong>How do I reproduce the benchmark results?</strong></summary>
192-
<br>
193-
Follow our detailed <a href="https://miromindai.github.io/MiroFlow/evaluation_overview/">Benchmarks Documentation</a> for step-by-step reproduction guides.
194-
</details>
195-
196-
<details>
197-
<summary><strong>Is there commercial support available?</strong></summary>
198-
<br>
199-
For commercial inquiries and enterprise support, please contact us through our <a href="https://miromind.ai/">website</a>.
200-
</details>
201-
202-
---
203-
204-
## 📖 References
90+
## References
20591

20692
If you find our work helpful, please consider citing:
20793

@@ -220,7 +106,7 @@ If you find our work helpful, please consider citing:
220106
@misc{2026miroflow,
221107
title={MiroFlow: A High-Performance Open-Source Research Agent Framework},
222108
author={MiroMind AI Team},
223-
howpublished={\url{https://github.com/MiroMindAI/MiroFlow}},
109+
howpublished={\url{https://github.com/MiroMindAI/miroflow}},
224110
year={2026}
225111
}
226112
```
@@ -229,13 +115,12 @@ If you find our work helpful, please consider citing:
229115

230116
<div align="center">
231117

232-
<a href="https://github.com/MiroMindAI/MiroFlow/graphs/contributors">
233-
<img src="https://contrib.rocks/image?repo=MiroMindAI/MiroFlow" />
118+
<a href="https://github.com/MiroMindAI/miroflow/graphs/contributors">
119+
<img src="https://contrib.rocks/image?repo=MiroMindAI/miroflow" />
234120
</a>
235121

236-
**Contributing**: [Issues](https://github.com/MiroMindAI/MiroFlow/issues) · [Pull Requests](https://github.com/MiroMindAI/MiroFlow/pulls) · [Discord](https://discord.com/invite/GPqEnkzQZd)
122+
**Contributing**: [Issues](https://github.com/MiroMindAI/miroflow/issues) · [Pull Requests](https://github.com/MiroMindAI/miroflow/pulls) · [Discord](https://discord.com/invite/GPqEnkzQZd)
237123

238124
**License**: Apache 2.0
239125

240126
</div>
241-

‎config/agent-browsecomp-gpt5-single-agent.yaml‎

Lines changed: 0 additions & 47 deletions
This file was deleted.

‎config/agent-gaia-validation-gpt5-multi-agent.yaml‎

Lines changed: 0 additions & 68 deletions
This file was deleted.

0 commit comments

Comments
 (0)