Palantir says Nvidia’s Nemotron Ultra beat frontier models after 24 hours
R_D · x · 2026-08-04
Palantir says Nvidia’s Nemotron Ultra performed “better than frontier” in its hands after just 24 hours, without any post-training.
- The claim comes from a quote shared by @jawwwn and amplified by @chetanp.
- According to the post, the team found the vanilla open-weight model outperformed frontier models on its customer tasks.
- The argument is that benchmark scores may miss the parts that matter for real applications, especially when developers can adapt open-weight models to their own workflows.
Related event: Palantir Says NVIDIA Model Beats Frontier Models in 24 Hours(3 posts)→
More from Companies & People
- Meta's Muse accused of mirroring OpenClaw's file structure; Nat Friedman denies copying — FinanceYF5 · 2026-09-22
- Alex Wang jokes his lifetime of being terminally online was training for this moment — alexandr_wang · 2026-09-22
- Vinod Khosla: Personal AI moats are data trust and getting tasks done — brucemacv · 2026-09-22
- Xiaomi's Lei Jun charted: from phones and EVs to AI, robotics, chips and satellite internet — bookwormengr · 2026-09-22
- SF post-training team hiring engineers obsessed with evals — sarahcat21 · 2026-09-22
- OpenAI says it won't share data, keeps all customer information private — MatthewChang · 2026-09-22