Developer
NVIDIA
17 indexed models; 4 currently meet the evidence threshold for the overall ranking. Together they have results from 18 evaluations.
Models by NVIDIA
| Model | Evals | Components | Rank | Release date |
|---|---|---|---|---|
| Nemotron 3 Ultra 550B A55B | 7 | 3/7 | 65 | 2026-06-03 |
| Nemotron 3 Super 120B A12B | 10 | 6/7 | 166 | 2026-03-10 |
| Nemotron 3 Nano 30B A3B | 8 | 4/7 | 191 | 2025-12-15 |
| Nvidia Nemotron Nano 9B V2 | 3 | 2/7 | 226 | 2025-08-12 |
| Llama 3.1 Nemotron Instruct 70B | 2 | 4/7 | — | — |
| Llama 3.1 Nemotron Ultra 253B V1 | 1 | 1/7 | — | 2025-04-07 |
| Llama 3.3 Nemotron Super 49B | 1 | 1/7 | — | 2025-03-16 |
| Llama3 Chatqa 1.5 70B | 1 | 1/7 | — | 2024-04-28 |
| Llama3 Chatqa 1.5 8B | 1 | 1/7 | — | 2024-04-28 |
| Nemotron 3 Nano Omni 30B A3B | 2 | 2/7 | — | — |
| Nemotron 3 Ultra | 1 | 2/7 | — | — |
| Nemotron 3.5 Lightning | 2 | 2/7 | — | 2026-08-11 |
| Nemotron Cascade 2 30B A3B | 1 | 1/7 | — | — |
| Nemotron Content Safety Reasoning 4B | 1 | 1/7 | — | — |
| Nvidia Nemotron 3 Nano 4B | 1 | 1/7 | — | — |
| NVIDIA Nemotron Nano 12B V2 | 1 | 1/7 | — | — |
| Nvidia Nemotron Nano 12B V2 Vl | 1 | 1/7 | — | — |
Evaluations covering NVIDIA models (18)
AA-Omniscience · AIMS Safety-Classifier Competence · Arena Factuality — Text Arena (factuality-only weighting) · BullshitBench v2 · ChineseSafe · Cisco AI Defense Rolling Single-Turn Leaderboard · DystopiaBench · Enkrypt AI Safety Leaderboard · Inkling-Small model card — FORTRESS · Inkling-Small model card — StrongREJECT · MT-JailBench CrescendoX · Olam Social Poker — Social Lie Rate · Pokee-Isaac model card — DTAP · RefusalBench · SM-Bench · SpeciEval · TAC · Vectara HHEM Factual Consistency