Sentient has launched an open-source AI search framework called Open Deep Search (ODS), claiming it outperforms established closed-source competitors such as GPT-4o and Perplexity AI. Announced on April 2, ODS is designed to be developer-friendly and emphasizes the preservation of developers' original intent through 'Loyal AI' models. Sentient utilizes fingerprinting technology to protect intellectual property while ensuring model openness, addressing the monetization challenges faced by open-source AI. ODS achieved a 75.3% accuracy on the 'Frames' benchmark, significantly higher than GPT-4o’s 50.5% and Perplexity Sonar’s 44.4%. Sentient’s co-founder emphasized the community-driven nature of AI development, asserting that open-source solutions outperform closed-source ones in both performance and user control. The launch coincides with an uptick in interest for open-source AI, as evidenced by over 1.8 million waitlist registrations ahead of the rollout. Research lead Sewoong Oh noted that the architecture of open-source models allows for competitive performance compared to proprietary systems, suggesting a pivotal shift in the landscape of AI development. In February, Sentient also completed a major NFT minting campaign, furthering its engagement in the blockchain space.

Source 🔗