Since DeepSeek challenged OpenAI two weeks ago, the open- vs. closed-source AI competition has shown no signs of stopping.
Also: Why Mark Zuckerberg wants to redefine open source so badly
Just two days after OpenAI announced Deep Research, a new AI agent within ChatGPT that can sift through online sources for you, its open-source counterpart has already emerged.
Hugging Face’s Deep Research
On Tuesday, Hugging Face Releasedan equivalent to the new feature. The alternative, blatantly called open Deep Research, uses OpenAI’s agentic framework and o1 model to navigate the web. The open alternative scored 55% accuracy in the General AI Assistants benchmarks (GAIA) (19459060) is a top assessment test of agents, compared with Deep Research’s 67%. It ranks first for open submissions. Hugging Face has acknowledged that the agent is still not a direct competitor to OpenAI. “Deep Research is a massive achievement and its open reproduction will take time,” The developer platform said in its blog titled ” “Freeing Our Search Agents.” “In particular, full parity will require improved browser use and interaction like OpenAI Operator is providing, i.e. beyond the current text-only web interaction we explore in this first step.”
Dumping Open Source for Proprietary rarely pays off: Better stick a fork into it.”
OpenAI Deep Research is based on a version of the company’s latest and most advanced reasoning models, o3, for which there is no open-source equivalent. According to OpenAI This model version outperformed top models in Humanity’s Last Exam, a new AI test released last week. It is also much more challenging than popular tests with a “new high” accuracy of nearly 27%.
Despite this, HLE’s designers are still adamant that their model is the best. Point outthat “contamination”is a potential: o3 has been evaluated after HLE’s release, so OpenAI had access its prompts. Hugging Face didn’t mention if it had tested Open Deep Research on HLE. The platform claims to be building “agents that view your screen and can act directly with mouse & keyboard.”
it’s free to test
Deep Research’s $200-per month price via ChatGPT may make it inaccessible for most. Open Deep Research offers a similar service for free. Live demo herewhich Hugging Face refers as a “simplified version” full agent.
Are ChatGPT Plus and Pro worth it? Here’s a comparison of the paid version and the free one
Hugging Face’s ability to create a model that is comparable to o1-preview in less than 24 hours shows the race that proprietary models are increasingly finding themselves in. Researchers at UC Berkeley created a model that was comparable to o1 preview in only 19 hours last month. DeepSeek has not revealed the exact timeframe for R1, its o1 competitor model. However, it is believed to be a lower-resource model in terms of both time and money.
Editorial Standards
