Nemotron 30B fine-tuned on Palantir's Ontology beats a model 18x its size
BrettKrieger12 · x · 2026-10-22
Palantir case data shows NVIDIA Nemotron models post-trained on its Ontology deliver staggering gains. In RFI classification, a fine-tuned Nemotron 3 Nano 4B hit 73.5% balanced accuracy at $0.1/1M tokens, beating Claude Sonnet 5.5 (66%) and GPT-6 Astra (62.5% at $20/1M); untuned it was only 53%. In Tampa General Hospital healthcare decision models, fine-tuned Nemotron 3.5 Lightning 30B reached 95.7% exact accuracy at 58ms latency, beating Claude Opus 5.5 (88.6%, 1,627ms) — a 30B model outperforming one 18x its size at a fraction of the cost and latency.
Related event: Palantir's Fine-Tuned Nemotron Beats Frontier Models at 1/20th the Cost(3 posts)→
More from Models
- Meta's Muse Spark Open: 816B-parameter open-weight model, 1M context, live on Together — togethercompute · 2026-10-22
- llama.cpp fork does live inference-time abliteration on any loaded model, no fine-tuning needed — AccountAntique9327 · 2026-10-22
- OpenAI Case Study: V7 Go Cuts Costs 78% with GPT-5.6 Luna — OpenAI · 2026-10-22
- Daniel Rock: New frontier models react far more aggressively to context instructions — danielrock · 2026-10-22
- Researchers credit OpenAI's Daybreak program with GPT-5.6 access for Chrome, GitLab exploits — Eric_Wallace_ · 2026-10-22
- Zeroing 6 neurons cuts Qwen3-8B sycophancy from 44% to 0%, interpretability charts show — Sauers_ · 2026-10-22