# Inherent, founded by DeepMind alumni, says its AI 'teammate' just outperformed Anthropic and OpenAI at replicating research

Published: 2026-08-22 · Updated: 2026-08-22 · Category: Research

Source page: https://www.sipply.dev/story/inherent-founded-by-deepmind-alumni-says-its-ai-teammate-c0a1e13f · 中文: https://www.sipply.dev/zh/story/inherent-founded-by-deepmind-alumni-says-its-ai-teammate-c0a1e13f

## Summary

Inherent, a new AI lab established by former DeepMind employees, has made headlines by claiming that its AI agent, Faraday, has surpassed the performance of larger models from industry giants Anthropic and OpenAI in the task of replicating scientific research findings. This accomplishment is particularly notable given that it was achieved using a smaller model, Qwen 3.6. The team at Inherent attributes this success to their innovative training methodology, which emphasizes developing a 'research taste' and enhancing collaborative instincts within the AI, allowing it to better understand and replicate complex scientific concepts.

## Key takeaways

- Inherent's AI, Faraday, has outperformed larger models from Anthropic and OpenAI.
- The success was achieved with a smaller model, Qwen 3.6, highlighting efficiency in AI training.
- The training approach focuses on 'research taste' and collaborative instincts, which may set a new standard in AI development.

## Why it matters

This development signifies a potential shift in AI research capabilities, demonstrating that smaller models can achieve significant results, challenging the notion that larger models are always superior. It opens up new avenues for efficiency and effectiveness in AI applications.

- **Product managers:** For product managers, this achievement suggests that investing in innovative training methodologies for AI can yield competitive advantages, even with smaller models, potentially reducing costs and time to market.
- **Developers:** Developers should take note of Inherent's approach, as it emphasizes the importance of training techniques that prioritize understanding and collaboration, which could lead to more effective AI systems in various applications.
- **Students and job seekers:** Students and job-seekers should recognize the significance of this breakthrough as it highlights the evolving landscape of AI research, suggesting that there are diverse paths to success in the field, including innovative approaches to model training.

## Sources

- [Inherent, founded by DeepMind alumni, says its AI 'teammate' just outperformed Anthropic and OpenAI at replicating research](https://techcrunch.com/2026/08/22/inherent-founded-by-deepmind-alumni-says-its-ai-teammate-just-outperformed-anthropic-and-openai-at-replicating-research) (TechCrunch AI, 2026-08-22)

## 中文

### 由DeepMind前员工创立的Inherent表示，其AI“队友”在复制研究方面超越了Anthropic和OpenAI

Inherent是一家由前DeepMind员工成立的新AI实验室，声称其AI代理Faraday在复制科学研究成果的任务中超越了行业巨头Anthropic和OpenAI的更大模型。这一成就尤其引人注目，因为它是使用较小的模型Qwen 3.6实现的。Inherent团队将这一成功归因于其创新的训练方法，强调培养“研究品味”和增强AI的协作本能，使其能够更好地理解和复制复杂的科学概念。

**要点：**

- Inherent的AI Faraday超越了Anthropic和OpenAI的更大模型。
- 这一成功是通过较小的模型Qwen 3.6实现的，突显了AI训练的效率。
- 训练方法侧重于“研究品味”和协作本能，可能为AI开发设定新的标准。

**为什么重要：** 这一发展标志着AI研究能力的潜在转变，表明较小的模型也能取得显著成果，挑战了更大模型总是更优的观念。这为AI应用的效率和有效性开辟了新的途径。

---
Sipply: Sip AI News Simply. https://www.sipply.dev